Laser-aided diagnostic of hydrogen isotope retention on the walls of the Globus-M2 tokamak

Possibility of using machine learning methods to reconstruct solid body parameters during laser-induced desorption analysis

“…When constructing the database of values {N i } , each of the listed parameters was varied in the following ranges: The thickness of the sample under study (its near-surface layer) L was taken equal to 1 μm, the parameters u m = 6.3 × 10 22 cm −3 and ν = 1.3 × 10 13 s −1 . The selected values of the sample parameters approximately corresponded to the conditions for conducting experiments on laser-induced desorption of hydrogen from layers re-deposited on the surface of tokamak tiles [29,33,35,36,43,44]. We emphasize that the choice of a fixed sample temperature in the calculations is a model approximation that ensures sample outgassing without the need to solve the problem of heat transfer or setting the temperature profile in the target, which significantly reduces the calculation time.…”

“…Its main goal is to demonstrate the possibilities of applying the proposed methods for determining the parameters of a solid body. More rigorous simulations of hydrogen thermal desorption from tokamak materials [29,33,36,44,[48][49][50] implies using more complex models that take into account the presence of several types of traps, their variable concentration and degree of filling over the sample depth, and others aspects, going far beyond the scope of this work.…”

“…To analyze data obtained in the framework of the LID-QMS, numerical codes, such as TMAP7 [34], are used to simulate dynamics of hygrogen thermal desorption from a sample [29,33,35,36], and recover the mass spectrometer time signals, by varying the parameters of a solid body. However, reconstructing the mass spectrometer signal is a multiparameter problem, and different sets of solid body parameters can lead to similar signal sweeps.…”

Dimensional effects in analysis of laser-induced-desorption diagnostics data

“…The diagnostics models are built around the set of desorption equations coupled with the heat equation describing the sample temperature dynamics under the pulsed laser irradiation. An example of such a code is TMAP7 [41] widely used for simulations of the tritium thermal desorption from the plasma-facing components during LID [5,[42][43][44]. Recently, the commercial packages, such as COMSOL Multiphysics also started to attract more attention for LID modeling [45,46].…”

“…The major assumption put into the desorption codes for modeling LID data is the application of onedimensional (1D) approximation for treating both the particle diffusion and heat conduction in the sample [32,42,44,47]. Although widely used for numerical analysis of data from conventional thermal desorption experiments [41,[48][49][50], in case of LID diagnostics the approximation can be violated, since the laser irradiation is always local.…”