Changes in Dislocation Punching Behavior Due to Hydrogen-Seeded Helium Bubble Growth in Tungsten

The accumulation of gas atoms in tungsten has been the topic of significant research as it is a critical issue for the durability of plasma-facing materials in fusion reactors. During fusion reactor operation, the expected high flux of gas atoms from the plasma to the W divertor will inevitably lead to their inclusion in the metal. These incident atoms primarily consist of Helium (He) and Hydrogen (H), including the H isotopes Deuterium (D) and Tritium (T), which act as fuel in the fusion reaction. Their inclusion in W can detrimentally alter its thermal and mechanical properties Chen et al. (2020); Iwakiri et al. (2000). Further, the implantation of gases under fusion conditions significantly modifies the surface morphology of the material, increasing the potential for release of high Z dust, impacting the energy efficiency of the plasma. Finally, there are concerns about the retention of radioisotopes such as T that impact the environmental handling of materials. The mitigation of these mechanisms is of paramount importance to ensuring the longevity and safety of reactor operation.

The inclusion of He or H (and its isotopes) in W has been studied extensively by the community. Computational work indicates that He inclusion leads to the formation of He bubbles which, during their growth in bulk W, generate glissile $1/2<$111$>$ dislocation loops which, at a certain size, emit from the bubble to a free surface Sefta et al. (2013); Iwakiri et al. (2000); Fikar et al. (2017); Sandoval et al. (2018). This process acts as a pressure release mechanism for the bubble and results in the creation of surface islands which act to progressively modify the W surface Hammond et al. (2020, 2019); Fikar et al. (2017); Fikar and Gröger (2015). Further, the bubble motion toward the surface as a result of progressive dislocation nucleation and punching or due to Frenkel pair induced transport Perez et al. (2017) can cause surface rupture leading to out-gassing of the He to the vacuum and further surface modification Sandoval et al. (2015); Ito et al. (2014). Through multi-scale modeling efforts, these mechanisms have been strongly correlated to the creation of the well-known W surface ‘fuzz,’ Dasgupta et al. (2019); Ueda et al. (2013, 2014); Wang et al. (2017a); Lasa et al. (2013, 2014); Chen et al. (2022) degrading the surface and producing nano-tendrils of W which emanate from its surface that could break off, impacting plasma performance. Somewhat counter to this are He bubbles growing at W GBs, as can be the case in polycrystalline W, due to the strong segregation of defects to a GB region El-Atwani et al. (2017, 2015); Bai et al. (2010); Uberuaga et al. (2015); Bai and Uberuaga (2013); Xiao et al. (2016). In this scenario the He bubble can instead act to transform the density of the GB phase as a pressure releasing mechanism rather than from forming surface altering dislocation loops, but this conclusion may depend on the character of the specific GB Hatton et al. (2024); Frolov et al. (2018).

Studies of H (D,T) inclusion in W commonly find the generation of surface blisters caused by the agglomeration of H gas into large subsurface bubbles due to their fast kinetics Lu et al. (2014); Liu et al. (2009a); Johnson and Carter (2010); Heinola and Ahlgren (2010); Frauenfelder (1969) and thermodynamic driving force to accumulate at W defects Johnson and Carter (2010); Duan et al. (2010); Liu et al. (2009b); Heinola and Ahlgren (2010); Ohsawa et al. (2010, 2012); Lu et al. (2014). This induces plastic strain at the surface Chen et al. (2020); Ye et al. (2003); Tanabe (2014); Lu et al. (2014) resulting in blisters, an effect seen in other materials subjected to a plasma which can, in some cases, result in surface rupture and dramatic out-gassing Hatton et al. (2019, 2020). It has also been suggested that GBs can have a trapping effect on H Yu et al. (2014); Wang et al. (2017b) which results in complex effects on the GB mobility and recrystallization of W Mathew et al. (2022) thereby impacting its mechanical properties Shi et al. (2023); Ding et al. (2022).

Experimental studies of a combined He and H flux indicate increased H retention in areas of the sample where He is present and an overall increase in H retention compared with a pure H exposure, implying some trapping enhancement caused by interactions between H and He Nishijima et al. (2005). This effect has been confirmed through computational work indicating that H preferentially decorates the periphery of the He bubble and is strongly bound to this position Bergstrom et al. (2017). Further, experimental studies which analyse surface damage as a result of combined gas fluxes provide some evidence that, compared to cases with only He or H fluxes, surface damage is significantly reduced Gao et al. (2014); Miyamoto et al. (2009); Alimov et al. (2009); Miyamoto et al. (2011). That is, in the combined case there is no evidence of surface blistering (as was the case with pure H implantation) nor surface fuzz (as with pure He implantation), indicating some proposed ‘cancelling’ effect when the two gasses are combined Gao et al. (2014); Miyamoto et al. (2009); Alimov et al. (2009); Miyamoto et al. (2011).

Previous computational studies of mixed He/H bubbles have concentrated on static properties, but did not shed light on the development of dislocation loops around the gas bubbles, an effect previously shown to be important when assessing surface W damage as a result of gas influx Sefta et al. (2013); Iwakiri et al. (2000); Fikar et al. (2017); Sandoval et al. (2018). The goal of the present study is precisely to investigate the effect that growing He:H bubbles have on the surrounding W matrix and the resulting character of the dislocation networks. This will allow the elucidation of the mechanisms which impact W surface damage in environments of combined He:H gas flux. More specifically, in section III.1 we use molecular dynamics (MD) to grow mixed He:H bubbles with a range of gas ratios. We analyse the behavior of the H in these bubbles and characterize the dislocation network which arises from these mixed gas bubbles. We then complete static calculations of dislocation loops which may result from He bubbles to understand how their formation energy is impacted by the presence of H. In section III.2, we analyse a specific case of an equal ratio He:H bubble close to the W surface to understand its behavior and impact on H retention and surface damage. Finally, in section IV we discuss the implications of the changes in SIA and dislocation behavior induced by the additional presence of H on W surface microstructural evolution.

This work uses the Molecular Dynamics (MD) code LAMMPS Thompson et al. (2022) with atomic interactions modeled using the W-H interatomic potential proposed by Gao et al. Wang et al. (2017c) which itself uses Marinica’s EAM potential Marinica et al. (2013) for W-W interactions. W-He and He-He interactions are modelled using the Juslin-Wirth potential set Juslin and Wirth (2013).

MD simulations of He:H bubble growth are conducted within a W supercell of size $127\times 127\times 60$ ų with $20$ Å of vacuum above and below the supercell to create two $<$100$>$ free surfaces in the z-direction. The supercell initially contains 64,000 atoms with periodic boundaries in each direction. A W atom is removed from the center of the supercell and 4 gas atoms are initially placed within this W vacancy site to seed the bubble with the desired He:H ratio, the structure is minimized, and the system is thermalized by running MD for 10 ps at 1000 K. MD is then conducted at 1000 K and a subsequent He or H is inserted randomly into a sphere of radius 1.4 Å centered on the initial W vacancy at a rate of 10¹¹ gas/s (or, every 10 ps). Due to H moving rapidly at these temperatures inside the bubble, it was necessary to use an MD timestep of 0.1 fs to avoid instability in the dynamics.

To understand the impact of H on the thermodynamic stability of different dislocation loop structures, static calculations of dislocation loop’s formation energies with and without the presence of H along their dislocation cores are also completed. Loops are created in a 150 Å cubic box with periodic boundary conditions in each direction initially containing 250,000 atoms. Loops with Burgers vectors along 1/2$<$111$>$ and $<$100$>$ are created of various sizes using a script which takes advantage of a multitude of modifiers within Ovito Stukowski (2010). The formation energy ($E_{f}$) is calculated using the standard formula,

Our atomistic simulations demonstrate that H can induce a change in the energetic favorability of dislocation loops formed from over-pressurized He bubbles. In particular, our work provides strong evidence that mixed He:H bubbles inhibit a bubble’s ability to generate glissile 1/2$<$111$>$ loops through a reduction in the formation energy of the competing, but sessile, $<$100$>$ loops. The importance of this finding lies in the effect that He bubbles are generally thought to have on the W surface morphology which is to cause the progressive roughening of the surface by periodic generation and punching of glissile loops Setyawan et al. (2023). These sessile loops, in contrast, cannot glide to the surface once produced and, in turn, cannot deposit their SIAs to the surface. The result of this is that interstitials stay attached to the bubble and its vicinity and can form a complex network of sessile dislocations as well as non-loop SIAs. The detailed behavior depends on the He:H ratio and is most evident in our simulations at mixtures close to an equal concentration of both gases.

This conclusion may be tempered by the fact that there is a well-known inconsistency with the interatomic potential used for this work, Marinica’s EAM potential Marinica et al. (2013) which is wrapped up into the Gao W-H potential set Wang et al. (2017c). The problem is this potential predicts a reversal in the trend in the stability of 1/2$<$111$>$ and $<$100$>$ dislocation loops compared to DFT calculations. That is, with the interatomic potential, $E_{f}^{<\text{100}>}$ $<$ $E_{f}^{1/2<\text{111}>}$, as seen in figure 3 (though only slightly). However, with DFT this trend is flipped such that $E_{f}^{<\text{100}>}$ $>$ $E_{f}^{1/2<\text{111}>}$, as can be seen in the shaded regions of figure 5 which describes the range of energetics described by DFT alongside a Moment Tensor Potential (MTP) developed specifically to capture the DFT dislocation thermodynamics, both taken from Ma et al. (2020).

To have some sense of how H might impact the stability of loops using DFT, we have shifted values of the computed DFT loop formation energies down by an amount corresponding to the decrease seen in figure 3 when that dislocation is saturated with H. That is, we consider a scenario in which we use the DFT energetics for the pure loops but combine that with the binding energy of H as determined with the potential. This shift shows that, especially for small dislocation loop sizes, $E_{f}^{<\text{100}>}$ $\simeq$ $E_{f}^{1/2<\text{111}>}$. This suggests that the presence of H may indeed impact the energetic favorability of each loop type in DFT and by extension, in experimental reality. Furthermore, there is some experimental evidence to suggest that surface damage is decreased as a result of a mixed flux of He and H to W Gao et al. (2014); Miyamoto et al. (2009); Alimov et al. (2009); Miyamoto et al. (2011) which is consistent with our findings using the interatomic potential. In the end, a primary finding of our MD simulations is that H more significantly impacts the stability of $<$100$>$ loops than it does 1/2$<$111$>$ loops, shifting the balance of the nature of loops punched from bubbles from glissile to sessile. Further DFT calculations may be needed to confirm these conclusions.

A trend that this potential does quantitatively capture is that of H trapping by both the He bubble and the associated dislocation loops. This process is thought to be responsible for the experimentally observed H blistering suppression caused by the blocking of the formation of H₂ molecules by the He bubble Zhou et al. (2010). Further, the presence of the He bubbles in the near surface region is shown to block the permeation of H deeper into W Markelj et al. (2017). However, in our atomistic study, we do find that if the mixed gas bubble approaches a critical distance from a free surface, the energetic cost of the glissile 1/2$<$111$>$ loop decreases and nucleates, pulling all W interstitials surrounding the bubble to the surface and releasing all H that was previously trapped by the dislocations. This is a new mechanism which must be considered when understanding fuel retention in W. A corollary observation from our simulations is that some H leaves the bubble to decorate the dislocation lines. This is observed mostly at high H concentration indicating that the He bubble is at first a stronger sink for H until some critical point and then the H prefers to decorate the dislocation loops, thus, dislocations can represent a pathway to drain bubbles of gas.

Finally, bubble growth rates chosen in computational studies are known to affect the behavior of the bubbles pressure profile over time such that bubbles grown at higher rate are higher in pressure than at a lower rate Sandoval et al. (2018); Hatton et al. (2022). Future studies could consider the impact of growth rate on the nature of punched dislocations in mixed-gas bubbles. However, we expect that the general conclusion, that H stabilizes $<$100$>$ loops and thus reduces interstitial transport via loop punching to free surfaces, is reliable as it is supported by our thermodynamic calculations.

Our atomistic study of the role of H on He bubble growth and loop punching behavior indicates that there is a significant impact to the dislocation generation and behavior in mixed He:H gas bubbles compared to pure He bubbles. H does this in multiple compounding ways. Firstly, H in low concentrations affects the ability for W interstitials to form dislocation loops leading to an increase in non-loop SIAs and therefore a reduction in the presence and length of dislocation loops. Secondly, at high H concentrations, H acts to inhibit the mechanisms of W Frenkel pair creation resulting in an overall decrease in SIAs. Finally, and most impactfully, we find a significant decrease in the generation of glissile 1/2$<$111$>$ dislocation loops when a bubble contains an equal mix of He and H. This is found to be due to a crossover in the energetic favorability of dislocation characters caused by H interactions with the dislocation cores. While further work is required to target these effects in experimental studies, we postulate our findings could have significant impact on the predicted damage evolution and fuel retention mechanisms in W in a fusion reactor environment.