Thermal-desorption spectroscopy (TDS) is widely used to characterize hydrogen-isotope retention in plasma-facing tungsten, but a measured spectrum does not uniquely identify the underlying microscopic defect configurations: different kinetic parameters can produce similar curves, while distinct vacancy-hydrogen configurations may have nearly the same binding energy. To account for these ambiguities, we formulate a Bayesian inversion that determines the range of detrapping-energy spectra consistent with a measured curve and reports all defect configurations allowed by those energies. Applied to a published deuterium-TDS curve of damaged tungsten, the method successfully resolves two detrapping-energy peaks centered at 1.58 [1.49, 1.70] and 1.97 [1.87, 2.08] eV, where each bracket contains 90% of the inferred probability. Across the four predamage conditions, the changing spectra can be described by the same two energy components while the lower-energy component loses relative weight as predamage temperature rises; both features also recur in an independent experiment. Tests on curves with known inputs show that peak temperatures are better determined than the detrapping energies behind them, and that the inferred energies still correspond to many vacancy-hydrogen configurations. The spectral evolution thus points to changing relative contributions from recurring trapping-energy families, without identifying a unique microscopic defect population.

Close

PrevNext

View figure in article

• Crystal defects
• Desorption
• Nuclear fusion
• Vacancies
• Bayesian methods
• Materials modeling
• Monte Carlo methods