In this section we consider some other cases, such as those involved in reaction R5 ignored above, and refocus some conclusions already made on other
reactions. We note here that so far the need to inves­tigate DD in irradiated metals was led mainly by the need to create multiscale modeling tools for predict­ing changes in mechanical properties due to irradia­tion. This is desirable for practical estimations in engineering support of real nuclear devices. How­ever, there is another consequence of dislocation activity that is related to microstructure changes that occur. While this may not be important in post­mortem experiments on irradiated materials, it can be important in real devices operating under irradiation. It is obvious that internal and external stresses can accumulate during irradiation ofcomplicated devices due to high temperature, radiation growth, swelling, and transition periods of operation, for example, shut down and restart. Creep is usually taken into account but microstructure changes due to dislocation activ­ity during irradiation are not. This activity affects the whole process of microstructure evolution and should therefore be taken into account in predicting the effects of irradiation. The validity of this state­ment is demonstrated by recent in-reactor straining experiments on some bcc and fcc metals and alloys.9 It is not possible at the moment to formulate unam­biguous conclusions and more experimental work will be necessary for this. Nevertheless, it is clear that dislocation activity during irradiation directly affects the radiation damage process. In the following section, we describe some mechanisms that can con­tribute to this at the atomic-scale level.

First, consider reaction R5, omitted in Section 1.12.4.2, for dislocation obstacles. Drag of glissile interstitial loops by moving edge dislocations was first observed in the MD modeling of bcc and fcc metals.89 It is well known that SIA clusters in the form of small perfect DLs exhibit thermally activated glide in the direction of their Burgers vector.90,91 It is characterized by very low activation energy ~0.01— 0.10 eV. An edge dislocation, having a long-range

elastic field, can interact with such clusters and, if bL is parallel to the dislocation glide plane, can drag or push them as it moves under applied stress/strain. The dynamics of this process have been investigated in detail and correlations between cluster and disloca­tion mobility analysed.92-94 Additional friction due to cluster drag reduces dislocation velocity, an effect that is stronger in fcc than bcc metals because offeatures of cluster structure.89 The maximum speed at which a dislocation can drag an SIA cluster is achieved by a compromise between dislocation-cluster interaction force and cluster friction and varies at T = 300 K from 180ms-1 in fcc Cu to >1000ms-1 in bcc Fe for loops containing a few tens of SIAs.89 An important consequence of this drag process is that a moving dislocation can sweep glissile clusters and transport them through the material.

Other reactions that may affect microstructure evolution involve both inclusion — and dislocation­like obstacles. The relevant reaction for the latter is denoted R3 in Table 1. In this case, an edge disloca­tion climbs and the formation of superjogs by defect absorption changes its structure and total line length. This changes its mobility and its cross-section for interaction with other defects, such as point defects, their clusters and impurities, and this in turn affects microstructure evolution. Reactions of type R2 can also be important for they change properties of obsta­cles. In thermal aging without stress, obstacles such as voids, SIA clusters and SFTs evolve towards their equilibrium low-energy state, that is voids/precipi — tates into faceted near-spherical shapes, SFTs into regular tetrahedron shape, and so on, whereas shear­ing creates interface steps on voids/precipitates and creates ledges on SFT faces. These surface features change the properties of the defects by putting them into a higher energy state.

Reaction R4 introduces another mechanism of mass transport, for the helical turn (representing the absorbed defects) can only extend or translate in the direction of its Burgers vector, that is, along the screw dislocation line. The case of dislocation-SFT interac­tion in a thin film considered in Section 1.12.4.2.1 has demonstrated that this may introduce completely new mechanisms. This effect can also play a role when a dislocation ends on an internal interface where it can cross-slip. Reaction R4 also orders the orientation of DLs left behind by a gliding screw dislocation for it changes bL of these loops to b of the dislocation, irrespective of their initial orientation.

Finally, we describe a case when several of the above mechanisms may have a significant effect on microstructure changes if operating at the same time on different defects. It is known from experimental studies6 that under neutron irradiation, Cu accumu­lates a high density of SFTs and this density saturates with dose at a high level (^1024m~3), close to con­ditions under which displacement cascades overlap. Taking into account that SIA clusters are necessarily accumulated in the system, this high saturation den­sity implies that annihilation reactions between the vacancy population in SFTs and SIA loops is sup­pressed. MD modeling in which an SIA cluster was placed between two SFTs about 10 nm apart and intersecting the loop glide cylinder has confirmed that annihilation reaction does not occur even after 50 ns at T< 900 K.95 The result is not surprising for each vacancy of an SFT is distributed over the four faces of the tetrahedron within the stacking faults.

However, simulations show that an annihilation reaction can be promoted by the involvement of a gliding dislocation. Two cases have been considered. In one, an edge dislocation under applied stress dragged a 1/2(110) SIA loop toward an SFT placed 7 nm below the dislocation slip plane and intersecting the glide cylinder of the loop. The overlapping frac­tions of SFT and dragged SIA loop annihilated by recombination. Different obstacle sizes (SFT from 45 to 61 vacancies and SIA loops from 37 to 91 SIAs) and geometries with different levels of overlap were simulated and recombination occurred in all cases. In the other situation, the same SFT and SIA loop 10-20 nm apart were placed on the slip plane of a screw dislocation. On approaching the obstacles, the dislocation absorbed a portion of each to form two helical turns (reaction R4). The turns of vacancy and interstitial character had opposite sign and the smal­ler was annihilated by recombination with part of the larger. On moving farther ahead, the dislocation released the unrecombined portion of the remaining helix to leave a small defect. This was usually part of the original SIA loop because a loop can be completely absorbed as a helical turn, whereas only a part of an SFT can be absorbed in this way. Thus, the overall result of interstitial loop drag under applied stress was annihilation of a significant part of both clusters by a reaction between helices of opposite signs.