Reply to Comment on ‘Deuterium supersaturated surface layer in tungsten: ion energy dependence’

We reply to the comment by Li et al (submitted to Nucl. Fusion with this response) on our recent paper Nishijima et al (2023 Nucl. Fusion 63 126003). In this response, we address the existence of an incident ion energy, E i, threshold for the deuterium (D) supersaturated surface layer (DSSL) formation with a newly conducted D plasma exposure experiment at E i ∼ 20 eV. It is also further demonstrated, based on new experiments where the ion flux, fluence, and sample temperature are scanned, that non-kinetic (ballistic) processes play a role in the DSSL formation and growth. In addition, the effect of impurities in our plasma is discussed also with a new analysis of the surface composition made after a D plasma exposure.

In our original paper [1], the formation of a deuterium (D) supersaturated surface layer (DSSL) in tungsten (W) was explored under D plasma exposure in the PISCES-A linear plasma device, as a function of the incident ion energy, E i , from ∼175 eV down to ∼45 eV.High-resolution transmission electron microscopy (TEM) observations confirmed the Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence.Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
DSSL formation, with high-density D nanobubbles in the layers, in the E i range investigated, even at the lowest E i ∼ 45 eV, which corresponds to a transferred energy from D to W, E D→W ∼ 1.9 eV.This E D→W value is much lower than the threshold displacement energy for W, i.e.E d ∼ 40 eV [2].As commented by Li et al, E i ∼ 45 eV is lower than the previously determined threshold, E i = 115 eV (E D→W ∼ 5 eV) in [3].We also proposed a DSSL formation mechanism, where E d could be effectively reduced by D interstitial atoms and induced lattice stress, as is made feasible according to the theories by others [4,5], and this proposed mechanism is described in our original paper.
In addressing Li's comment, it is conceded that an E i threshold for DSSL formation was not reported in our original paper, but this does not mean there is no energy threshold, and it was not intended that this message be conveyed, should it have been a source of misunderstanding by the commenting Authors.
Independently of the comment by Li et al further work has now been done to address an energy threshold for the DSSL formation, by a further experiment in PISCES-A, where a W sample was exposed to D plasma at E i ∼ 20 eV (E D→W ∼ 0.86 eV), less than the lowest E i ∼ 45 eV in the original paper.Other plasma exposure parameters were identical to those in [1]: sample temperature, T s , ∼ 423 K, ion flux, Γ i , ∼ 1.2 × 10 21 m −2 s −1 , and fluence, Φ i , ∼ 3.0 × 10 24 m −2 .After the plasma exposure, a thin lamella specimen was first created from the W sample using a focused ion beamscanning electron microscopy system (HITACHI NB5000), and then observed with a TEM device (JEOL JEM-2800) at NIFS, Japan.
Figure 1(b) shows a cross-sectional TEM image of the subsurface region of the W sample exposed at E i ∼ 20 eV, together with the (a) unexposed and (c) E i ∼ 45 eV samples [1] for comparison.At E i ∼ 20 eV, no D nanobubbles can be seen, indicating that a DSSL did not form.This evidence demonstrates there is an E i threshold for DSSL formation between 20 and 45 eV.With the new data at E i ∼ 20 eV, figure 2(b) of [1] is updated by figure 2 here: the E i dependence of the measured DSSL thickness, ∆t DSSL .Note that the 0.1× peak concentration depth is close to the maximum implantation depth (see figure 2(a) in [1]).
Li et al consider only kinetic processes for displacement of lattice W atoms; ∆t DSSL depends only on E i (E D→W ).After the original publication, we have further investigated the dependence of ∆t DSSL on other plasma exposure parameters, Γ i , Φ i , and T s , at E i ∼ 75 eV.Figure 3 displays cross-sectional TEM images of DSSLs formed in W as a function of (a)-(c) Γ i , (d)-(f ) Φ i , and (g)-(i) T s .In each parameter scan, the other parameters were kept constant, as indicated in the figure.This new experiment reveals that ∆t DSSL increases with increasing both Γ i and Φ i , while it is nearly constant against T s in the narrow range of ∼353-523 K. Since the implantation range does not depend on Γ i and Φ i , arguments by Li et al cannot explain the observed increase of ∆t DSSL as Γ i and Φ i increase.Furthermore, this new evidence also means that non-kinetic (ballistic) processes also play a role in DSSL formation and growth.With a higher Γ i , more interstitial D atoms exist in the implantation range, which may more effectively reduce E d [4] and also induce stresses in the W matrix.This can promote the DSSL formation, as discussed in our original paper.Damage appears to accumulate with increasing Φ i , and the layer grows.Interestingly, interbubble fracture [8], indicated by circles in figure 3(f ), which can result from the evolution of high-density D nanobubbles, occurs at the highest Φ i .In view of this new experimental evidence, the E i (E D→W ) threshold for DSSL formation is now expected to be also influenced by Γ i and Φ i .We should note that, while the T s dependence is not observed in the present experiment, T s may also affect the DSSL threshold outside the range scanned here and with different plasma exposure parameters.
Figure 4 summarizes the measured ∆t DSSL as a function of (a) Γ i and (b) Φ i .Here, it is important to note that, although ∆t DSSL is in good agreement with the 0.1× peak concentration depth in figure 2, this is valid only for the quoted plasma exposure condition.That is why the ∆t DSSL value from [3], with a different plasma exposure condition (T s = 300 K, Γ i = 9.9 × 10 19 m −2 s −1 , and Φ i = 6 × 10 24 m −2 ), deviates from the curve.We should also note that the bulk W density was assumed for nuclear reaction analysis to derive ∆t DSSL in [3], which is thought to underestimate ∆t DSSL as highdensity D nanobubbles reduce the W density in a DSSL.Gao et al mentioned in the comment that the excellent agreement of the SSL thickness between experiments and SDTrimSP simulations [3] seems to validate their energy threshold of E D→W = 5 eV.However, we believe that the excellent agreement is a coincidence, since non-kinetic processes related to Γ i and Φ i are not taken into account.In addition, we can find no compelling evidence for choosing a threshold of 1% of the peak value of the number of collisions per depth interval with E D→W above a certain energy, as was chosen to determine the ∆t DSSL in the simulations [3].
One might also question the uniqueness of SSL formation to hydrogen isotopes (HIs).For instance, it is well known that low-energy helium (He) plasma exposure of W also induces damage in the near-surface region due to the formation of a high-density He nanobubble layer, that results from He precipitation not far from the implantation range under conditions where the incident He ions do not have sufficient energy for atomic displacement of W. The reader is referred to [9] for details, and we note that this is another example of damage creation due to non-kinetic processes that highlights the fact that the SSL formation process might not be limited to the effects of HIs.
Lastly, as pointed out by Li's comment, and discussed in our original paper, trace plasma impurities of mainly carbon (C), can exist in the PISCES-A plasma, just as in many other plasma devices.The effect of impurities has also now been further studied, independently of the comment by Li et al.Energy-dispersive x-ray micro-analysis has been performed on selected W samples, that were exposed to D plasma in PISCES-A in this series of experiments.C impurities of ∼20 at.% were detected on both exposed and unexposed (covered by a sample holding cap during the plasma exposure) surfaces.This fact indicates that C impurities detected on the different surface regions originate mostly from air exposure after the D plasma exposure.If C were deposited or implanted into the surface, these differing regions, protected and exposed, would almost certainly show differing C concentrations.Nevertheless, a C impurity influence cannot be completely ruled out.
Contrary to our view, but on the assumption that C does play a role, we point out that the transferred energy from C to W at E i ∼ 45 eV is only ∼10 eV, still considerably less than the E d ∼ 40 eV needed for displacement of W, again meaning that additional mechanisms would still be required to effectively reduce E d , e.g.D interstitial atoms [4] and stresses [5], as was discussed in our original paper.In addition, the implantation depth of C in W is even shallower than that of D in W such that the observed DSSL thicknesses in experiment would be inconsistent with the C in W implantation range.We note that the experimentally observed DSSL thicknesses are more consistent with the D in W implantation range (notwithstanding the new dependencies also on Γ i and Φ i ).
Moreover, speculation by Li et al regarding chemical reactions of D with impurities causing the DSSL formation, we think, is unlikely.If chemical reactions of D dominate the DSSL formation in our experiments, a DSSL should also form at E i ∼ 20 eV.Yet, this is not what is observed by experiment, as demonstrated in figure 1.
In conclusion, our latest experiments have confirmed no DSSL formation at E i ∼ 20 eV, meaning there exists an E i (E D→W ) threshold between 20 (0.86) and 45 (1.9) eV for the DSSL formation under the present plasma conditions.Furthermore, the Γ i and Φ i scan experiments have revealed that non-kinetic processes also play a role in the DSSL formation and growth.In other words, the E i (E D→W ) threshold depends on Γ i and Φ i , as well as possibly on T s .

Figure 1 .
Figure 1.Cross-sectional TEM images of W samples: (a) unexposed, (b) exposed at E i ∼ 20 eV, and (c) ∼45 eV.The scale is the same for all the images.Reproduced from [1].© 2023 The Author(s).Published by IOP Publishing Ltd on behalf of the IAEA.All.

Figure 2 .
Figure 2. E i dependence of the measured ∆t DSSL (b) of [1] is updated with the new data at E i ∼ 20 eV.The solid line shows the implantation depth at the 0.1× peak concentration, calculated with SDTrimSP [6, 7].

Figure 3 .
Figure 3. Cross-sectional TEM images of DSSLs in W formed as a function of (a)-(c) Γ i , (d)-(f ) Φ i , and (g)-(i) Ts.The scale is the same for all the images.Note that the images in (b), (e), (h) are identical (reference condition).Reproduced from [1].© 2023 The Author(s).Published by IOP Publishing Ltd on behalf of the IAEA.All.

Figure 4 .
Figure 4. Measured ∆t DSSL vs. (a) Γ i and (b) Φ i .Other plasma parameters are indicated inside the figures.

Figure 5 .
Figure 5. (a) Under focused and (b) over focused bright field images of a DSSL.White dots in (a) are seen as black dots in (b).Representative cavities are pointed by arrows.