Modelling the Performance of Tritium Process Monitors from First Principles

TL;DR

This work presents a first-principles computational model of a one-liter ionization-chamber Tritium monitor, solving the internal electrostatics and particle dynamics to predict the detector current under ideal and non-ideal conditions. It combines a Jacobi-based electric-field solver, a Particle-In-Cell treatment for ion interactions, a Debye-length–driven spatial discretization, and a five-step time evolution (deposition, interaction, recombination, position evolution, and detection) to capture beta-induced ion production, transport, and loss. The model successfully reproduces the linear saturation current and the voltage-, pressure-, and activity-dependent deviations observed in experiments, providing quantitative insights and a recombination coefficient for argon that can be tuned to data. This framework enables design optimization for higher-activity monitors and can be extended to additional detector geometries (e.g., 20 cc wire cage designs) to guide future detector development.

Abstract

Ionization chamber-based, in-line tritium process monitors play an important part in determining the behavior of a tritium system. The one-liter detection volume monitor has been characterized well through experiment and to respond linearly to tritium concentrations for the range of $1 μCi/m^3$ to $1 Ci/m^3$. Additionally, it has been shown to behave nonlinearly for low voltage on the central anode and low pressure of the carrier gas. A computational model was developed from first principles that successfully describes the behavior of the one-liter monitor for each of these regimes. Predictions from the model are compared to previously-collected experimental data in order to determine validity and tune the model. The model will be expanded to incorporate additional detector geometries and designs in the future.

Figures (6)

• Figure 1: A radial cross-section of the electric potential (in volts) as a function of position in the chamber. Overlaid on the graph is the direction (but not the magnitude) of the electric field, which lies perpendicular to the equipotential lines.
• Figure 2: Average current seen by the chamber throughout a series of timesteps. Within five seconds, the average current calculated by the model converges to a stable value, indicating to the program that no further iterations are required.
• Figure 3: At 50 evenly spaced concentrations between 1 and 50 $\mu Ci/m3$, the simulation is run for a fixed gas pressure of 760 Torr and anode voltage of -100 V. The carrier gas used for this series of trials was nitrogen. The blue line is the predicted response of the process monitor according to the model. The points correspond to previous experimental data ProcessMonitors collected under ideal conditions.
• Figure 4: Another 15 evenly spaced concentrations are selected between 1$\mu Ci/m3$ and 104 $mCi/m3$. The same conditions of -100V and 760 Torr nitrogen are used for this trial. Again, the blue line is the prediction of our model, while the red points correspond to previously measured behavior of the process monitor ProcessMonitors. The response of the detector remains in the ideal regime as the electric field remains sufficient to keep ion densities low at higher tritium concentrations.
• Figure 5: At -1, -3, -5, -8, -10, -30, -50, -80, -100, -300, and -500 Volts on the central anode, the simulation is run for a fixed argon carrier gas pressure of 760 Torr at the three fixed concentrations of activity: 9.7 $mCi/m^3$, 105 $mCi/m^3$, and 1.74 $Ci/m^3$. The selected recombination coefficient, $\rho = 3\times10^{-6} \frac{cm^3}{ion \cdot sec}$, was selected to align the model to experimental data. The red points represent previously collected experimental data ProcessMonitors, while the colored curves correspond to the predictions made by the program. As with experiment, it is found that the magnitude of the voltage required to remain in the ideal regime increases with the true concentration of activity inside of the chamber.