08.05.2013 Views

Volumen II - SAM

Volumen II - SAM

Volumen II - SAM

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Table 1 reports the first few processes involving the 2I cluster. (D/C)n stands for a pair of<br />

dumbbells/crowdions located at n th nearest neighbors, whereas the star indicates variant forms. In bold is the<br />

proposed activation energy, that for both models entails a jump at 1nn with rotation of the common<br />

dumbbells' axis ( e.g. (000) → 1/2(111) ). However, the (optimum) two-steps mechanism that<br />

accomplishes the jump is model dependent. For EAM, one of the dumbbells moves first to a 3nn position,<br />

then the other follows, whereas for ED the first step involves the 2nn; this is shown in figure 3, where one<br />

should interpret, e.g., the 2I EAM sequence as ( ab → ab1 → a2b1 ).<br />

Due to the number of possible structures, the precise description of the 3I case becomes more intricate,<br />

though qualitative arguments can still be given. For EAM the relevant structure consists of three <br />

(distorted) crowdions; the one located at the cluster tail is the least extended and rests misaligned with the<br />

other two. The latter feature entails that the jump is (slightly) sense dependent, being 0.22 eV the activation<br />

energy for motion along the row. On the other hand, the defect may exchange directions<br />

through an immobile configuration, involving a barrier of 0.23 eV.<br />

The situation regarding 3I ED is better described in terms of dumbbells, though motion still happens along<br />

the rows and involves a sequence of individual jumps. The behavior resembles somewhat the EAM<br />

case in that the dumbbell more advanced' may eventually lock motion. The sequence comprises 4 steps, the<br />

2 nd (and 3 rd ) being the critical one, with a barrier of 0.31 eV.<br />

Regarding the 4I cluster, as stated for the MD simulations, migration is tied to the row. Being the<br />

Burgers vector of the ED rhombus -type in nature, a sizable amount of activation is needed; it was<br />

found that the activated form is reached in about half a dozen steps with 0.82 eV as overall barrier.<br />

Other processes, not leading to migration, also compete, e.g. a non-dissociative jump of the dumbbells<br />

located across the longest rhombus diagonal. Applied to the EAM case, the method turns out to be very<br />

inefficient; in particular no barrier for rotation could be determined. On the other hand, the overall barrier for<br />

motion was computed to be 0.33 eV.<br />

4. CONCLUSIONS<br />

Though the main characteristics of migration of SIA clusters beyond some 7 units, are most likely dominated<br />

by lattice structure, we have shown that the behavior of clusters of up to 4 SIA may be influenced by the<br />

kind of interaction model chosen. This is true at the level of the size of the quantities involved, such as<br />

activation energy, but also when considering the relevant micromechanisms, e.g., for the central force EAM,<br />

extended structures are clearly preferred..<br />

We have also shown the usefulness of a (static) transition state finding technique, the Monomer, at filling in<br />

details where MD can hardly do, and also at providing comparable activation energies at a fraction of the<br />

computational cost. Besides some points discussed below, a drawback of the static technique (at least in the<br />

current naive implementation) is the increased inefficiency with cluster size. Perhaps a combined strategy,<br />

with MD providing initial configurations and search directions, may alleviate this pitfall.<br />

Overall, the agreement between Monomer and MD is reasonable. However, even though inherent statistical<br />

errors might be invoked where disagreement is more apparent, one may wonder, for instance, why for the<br />

simplest case of the 1I cluster the static technique obtains values systematically larger. Part of the answer<br />

may lie in entropic contributions present in MD, that relatively stabilize easier migrating structures beyond<br />

their naive Boltzmann weights [19]. Entropy effects were also reported in [21] to stabilize immobile cluster<br />

configurations in Fe; however, noting the rather high temperatures involved in this effect, one may question<br />

their importance to cascade damage conditions. In any case, the link between the static results and the<br />

dynamic ones is not straightforward. The former can be refined by computing vibrational spectra and attack<br />

frequencies in the spirit of transition state theory, though at the expense of new questions on their reliability<br />

with current potentials. The central difficulty however is that these clusters, complex objects after all, may<br />

not suit the picture of random walkers. In fact, under certain conditions not even the single SIA does, as<br />

found in recent models for V and W [22,23]. Nevertheless, considerations such as these are too fine details to<br />

be accounted for within the higher hierarchy theories of radiation damage evolution, which simulations like<br />

the present ones try to inform.<br />

ACKNOWLEGMENTS<br />

Partial support from CONICET-PICT 5062 is gratefully acknowledged. Thanks also to Dr. G. Simonelli for<br />

sharing her results during the initial stages of the current research.<br />

1381

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!