Depending on the complexity of the microstructure, internal interfaces such as grain boundaries, twins,

and lath and packet boundaries (in ferritic/martensitic steels) can provide a significant sink in the material for point defects. As such, they may play a significant role in radiation-induced microstructural evolution.

For example, the effect of grain size on austenitic stainless steels was observed as early as 1972.112-114 The swelling effect was more closely associated with damage accumulation than damage production, but current understanding of the role of mobile inter­stitial clusters has provided a link to damage produc­tion as well (Singh and coworkers115 and Chapter

Vacancy cluster size (4-NN)

To date, there have been a limited number of stud­ies carried out to investigate whether and how primary damage formation would be altered in nanograined metals,116-121 and quite strong effects have been observed.116 The work from Stoller and coworkers122 will be used here to illustrate the phenomenon because the results of that study can be directly compared with the existing single crystal database that has been dis­cussed above. A sufficient number of simulations were carried out at cascade energies of 10 and 20keV and temperatures of 100 and 600 K to obtain a statistically significant comparison. The results demonstrate that the creation of primary radiation damage can be sub­stantially different in nanograined material due to the influence of nearby grain boundaries.

To create the nanocrystalline structure, grain nucleation sites were chosen, and the grains were filled using a Voronoi technique.123 A 2 x 2 x 2 lattice parameter face-centered cubic (fcc) unit cell system was used to obtain the grain nucleation sites, resulting in 32 grains in the final sample. Each Voronoi poly­hedron was filled with atoms placed on a regular bcc iron crystalline lattice, with the lattice orientation randomly selected. Grain boundaries occur natu­rally when the atomic plains in adjacent polyhedra impinge on one another, and overlapping atoms at the grain boundaries were removed. The final system was periodic and had an average grain size of 10 nm, system box length of 28.3 nm, and contained roughly 1.87 million atoms. More details on the procedure can

be found in Stoller and coworkers.122 The system was equilibrated for over 200 ps including a heat treat­ment up to 600 K. Figure 29 illustrates a typical grain structure with each grain shown in a different color. The approximate sizes of 5 and 20 keV cascades have been projected on to the face of the simulation cell.

MD cascade simulations were carried out in the same manner discussed above, although the analysis was somewhat more difficult due to the need to differentiate cascade-produced defects from the defect structure associated with the grain boundaries. It was common for the cascade volume to cross from one grain to another.1 2 The number of vacancies and
interstitials surviving in the nanograin simulations is compared to the single crystal results in Figure 30. A wider range of cascade energies is included in Figure 30(a) to show the trend in the single crystal data, while Figure 30(b) highlights the differences at the temperature and energy of the nanograin simulations. Mean values are indicated by the sym­bols in Figure 30(a) and the height of the bars in Figure 30(b), and the error bars indicate the standard error in both cases. Similar to the case for surface — influenced cascades, the number of surviving vacan­cies and interstitials is not the same for cascades in nanograined material. The number of vacancies sur­viving in the nanograined material is similar to the single crystal data for 10 keV cascades, but higher at 20 keV. Much lower interstitial survival is observed in nanograined material under all conditions.

Consistent with the overall reduction in interstitial survival shown in Figure 30(a), the number of inter­stiials in clusters is dramatically reduced in nano­grained material for all the conditions examined. As the number of surviving point defects, particularly interstitials, is so strongly reduced in the nanograin material, it is helpful to compare the fraction of defects in clusters in addition to the absolute number. Such a comparison is shown in Figure 31 where the fractions of surviving interstitials and vacancies contained in clusters in both nanograined and single crystal iron are compared for all the conditions simulated. The relative change in the clustering fraction is somewhat

less than the change in the total number of defects in clusters, but are still substantial for interstitial defects. Notably, the temperature dependence of clustering between 100 and 600 K observed in the single crystal 20 keV cascades is reversed in nanograined material. Between 100 and 600 K, the fraction of interstitials in clusters increases for single crystal iron but decreases for nanograined iron. Conversely, the vacancy cluster fraction decreases for single crystal iron and increases for nanograined iron.

Although the range of this study was limited in temperature and cascade energy, the results have demonstrated a strong influence of microstructural length scale (grain size) on primary radiation damage production in iron. Both the effects and the mechan­isms appear to be consistent with previous work in nickel,116,120 in which very efficient transport of interstitial defects to the grain boundaries was observed. In both iron and nickel, this leads to an asymmetry in point defect survival. Many more vacancies than interstitials survive at the end of the cascade event in nanograined material while equal numbers of these two types of point defects survive in single grain material. Similar to single crystal iron,59,64 few of the vacancies have collapsed into compact clusters on the MD timescale. The vacancy clusters in both single and nanograined iron tend to be loose 3D aggregates of vacancies bound at the first and second NN distances as shown above in Figure 20. The size distribution of such vacancy clusters was not significantly different between the single and nanograin material. In contrast, the
interstitial cluster size distribution was altered in the nanograined iron, with the number of large clus­ters substantially reduced. There appears to be both a reduction in the number of large interstitial clusters formed directly in the cascade and less coalescence of small mobile interstitial clusters as the latter are being transported to the grain boundaries.

The changes in defect survival observed in these simulations are qualitatively consistent with the lim­ited available experimental observations.117-119 For example, Rose and coworkers117 carried out room — temperature ion irradiation experiments of Pd and ZrO2 with grain sizes in the range of 10-300 nm, and observed a systematic reduction in the number of visible defects produced. Chimi and coworkers118 measured the resistivity of ion irradiated gold speci­mens following ion irradiation and found that resis­tivity changes were lower in nanograined material after room-temperature irradiation. However, they observed an increased change in nanograined material following irradiation at 15 K. The low-temperature results could be related to the accumulation of excess vacancy defects as they would be immobile at 15 K.