Isotopic correlation for ²⁴²Pu composition prediction: Multivariate regresssion approach

Area of Science:

• Nuclear Chemistry
• Isotope Analysis
• Radiochemistry

Background:

• Accurate prediction of plutonium isotopic abundances is crucial for nuclear forensics and safeguards.
• Existing empirical methods for predicting plutonium-242 (²⁴²Pu) abundance have limitations in accuracy.
• Understanding isotopic correlations, particularly the role of plutonium-238 (²³⁸Pu), is key to improving predictive models.

Purpose of the Study:

• To evaluate and compare multivariate regression techniques for predicting ²⁴²Pu isotopic abundance.
• To determine the significance of ²³⁸Pu abundance in the prediction of ²⁴²Pu.
• To assess the performance of multiple linear regression (MLR) against principal component regression (PCR), partial least squares regression (PLSR), and established empirical methods.

Main Methods:

• Application of multivariate calibration techniques: multiple linear regression (MLR), principal component regression (PCR), and partial least squares regression (PLSR).
• Utilized atom percent abundances of ²³⁸Pu, ²³⁹Pu, ²⁴⁰Pu, and ²⁴¹Pu as input variables.
• Comparative analysis of MLR predictions against seven previously reported empirical methods, including the Bignan correlation.

Main Results:

• MLR demonstrated superior prediction capability for ²⁴²Pu abundance compared to PCR and PLSR.
• The abundance of ²³⁸Pu was found to have a small but significant effect on ²⁴²Pu prediction accuracy, particularly for achieving <0.5% error.
• The MLR model significantly outperformed all seven empirical methods, with the Bignan correlation (using ²³⁸Pu) showing the best performance among them.

Conclusions:

• MLR is a highly effective method for predicting ²⁴²Pu isotopic abundance.
• Incorporating ²³⁸Pu isotopic data into predictive models is essential for achieving high accuracy in ²⁴²Pu abundance estimations.
• The study confirms a small but important correlation between the production of ²³⁸Pu and ²⁴²Pu.