Removal of spallation-induced tritium from silicon through diffusion

Abstract

Tritium, predominantly produced through spallation reactions caused by cosmic ray interactions, is a significant radioactive background for silicon-based rare event detection experiments, such as dark matter searches. We have investigated the feasibility of removing cosmogenic tritium from high-purity silicon intended for use in low-background experiments. We demonstrate that significant tritium removal is possible through diffusion by subjecting silicon to high-temperature (> 400C) baking. Using an analytical model for the de-trapping and diffusion of tritium in silicon, our measurements indicate that cosmogenic tritium diffusion constants are comparable to previous measurements of thermally-introduced tritium, with complete de-trapping and removal achievable above 750C. This approach has the potential to alleviate the stringent constraints of cosmic ray exposure prior to device fabrication and significantly reduce the cosmogenic tritium backgrounds of silicon-based detectors for next-generation rare event searches.

Figures (10)

• Figure 1: Diced wafer loaded in quartz sample boat.
• Figure 2: Bubbler change times at each temperature
• Figure 3: Time profile of tritium outgassing from silicon wafers at different temperatures. The black bin edges mark the start and end of each bubbler change while the central marker shows the measured activity and uncertainty for each bubbler. Results of the final bake at 1000 are shown in green. The blue and red markers show the best-fit result to a diffusion-only and diffusion + de-trapping model (see Section \ref{['sec:fitting']} for details).
• Figure 4: Fraction of tritium removed from the silicon wafers at different baking temperatures. The black data points show the experimentally measured fraction removed compared to the total (data point at 1000 assumes complete removal). The red data points are the results from the fit of the time profile to the diffusion and de-trapping model, including both the free and de-trapped contributions.
• Figure 5: Comparison of measured cumulative outgassed activity to the best fit models. Fits were performed on the independent bubbler measurements (see Figure \ref{['fig:3H evo profiles']}) but are shown in terms of the cumulative activity for easier visualization of trends. Blue: Fit to simple diffusion model. Red: Fit to model with diffusion and de-trapping. Note that the horizontal axes of the plots span different durations.