Unfaulting of faulted Frank loops V: experimental observations

Having examined the crystallography of unfaulting reactions in alumina and spinel, it is interesting now to compare and contrast what is experimentally observed so far as dislocation loop evolution in irra­diated Al2O3 versus MgAl2O4 is concerned. First, it is observed that the 1/3 [0001] (0001) basal loops and 1/3(1010){1010} prismatic loops readily unfault under irradiation, by the reactions shown in eqns [7,8] and [9,10], respectively.6 These reactions occur when the loops reach ^50 nm diameter,10 and each reaction produces an unfaulted loop with a 1/3(1011) perfect Burgers vector. Once formed, these unfaulted loops grow without bound until they intersect other growing dislocation loops, ultimately forming a dis­location network. Such a dislocation network in neu­tron irradiated Al2O3 is shown in Figure 6.

Once the dislocation network in irradiated alu­mina is formed, it has been demonstrated that the product dislocations within the network are free to climb6 The continuous climb of network disloca­tions in Al2O3 provides unsaturable sinks for Al and O interstitials arriving in stoichiometric proportions. All the conditions for a substantial supersaturation of vacancies are now in place. Al and O interstitials are readily absorbed at network dislocations, leaving behind numerous unpaired Al and O vacancies in the lattice. These unpaired vacancies inevitably con­dense to form voids. Under these conditions, void swelling must be the anticipated radiation response of the material.

Contrast the evolution described above for alumina to the observed microstructural evolution in spinel. The predominant 1/4 (110) {110}

image388

Figure 6 Weak-beam dark-field transmission electron micrograph showing the dislocation network formed in Al2O3 following neutron irradiation at 1015 K to a fluence of 3 x 1025nm~2 (~3dpa). Reproduced from Clinard, F. W., Jr.; Hurley, G. F.; etal. J. Nucl. Mater. 1982, 108/109, 655-670.

dislocations in spinel do not unfault under most experimental conditions tested to date.6 (Kinoshita eta/.51 observed unfaulted 1/2 (110) {110} perfect loops in MgAl2O4 single crystals following neutron irradiations in the JOYO fast breeder test reactor (fast neutron fluences up to 6.5 x 1025nm~2 (equivalent to 6.5 dpa), and temperatures between 673 and 873 K). Kinoshita eta/.51 also proposed a growth pro­cess of loops in spinel as follows: 1/6 [111] (111) 1/4 [110] (111) 1/4 [110] (101) 1/4 [110] (110) 1/2 [110] (110). Notice that this sequence ends in an unfaulted, perfect interstitial loop. This final loop configuration should be a good sink for interstitials, thus promoting a supersaturation of vacancies in the lattice. However, in neutron irradiations of MgAl2O4 single crystals in the fast flux test facility (FFTF), no evidence for 1/2 [110] (110) perfect dislocations was found, for neutron fluences ranging from 2.2 x 1026 to 2.17 x 1027nm~2 (equivalent to 22-217dpa) in the temperature range 658-1023 K.51 Therefore, the proposed progression of spinel interstitial loop characteristics described above has, to date, been confirmed only under the JOYO irradiation condi­tions reported by Kinoshita et a/.51) According to Clinard eta/.:6

Persistence of the 1/4 (110) {110} stacking fault amounts to a failure of a 1 /4(112) partial dislocation to nucleate, sweep across the loop plane, and so remove the fault.

The reason for this failure is paradoxical. Apparently, stacking fault energy cannot be the reason. The stack­ing fault energy estimate for 1/4 (110) {110} stacking faults in spinel, 180 mJ m~2,18 is similar to the energy estimates for 1/3 [0001] (0001) and 1/3(10І0){10І0} stacking faults in alumina (320 and 750 mJ m~2, respectively).10 Therefore, there seems to be a rea­sonable ‘driving force’ available to favor unfaulting of 1/4 (110) {110} stacking faults in spinel. Perhaps the explanation is simply that the magnitude of the par­tial shear vector required to unfault the faulted loops is prohibitively large. In spinel, the magnitude of the unfaulting 1/4(112) vector is ~5A, compared with the 1/3 [0001] (4.32 A) and 1/3(1010) (2.74 A) unfaulting vectors in alumina.

Whatever the reason, spinel 1/4 (110) {110} stacking faults do not unfault, and this leads to void swelling resistance and impressive inherent radiation tolerance in spinel compared alumina. Hobbs and Clinard summarize the situation as follows:

The absence of void swelling ( in spinel ) can be attributed to the failure of the loops to unfault and develop into dislocation networks; they therefore remain less than perfect interstitial sinks since the energy per added interstitial never drops below the fault energy. Vacancy-interstitial recombination thus remains the dominant mode ofdefect accommo­dation, and saturating defect kinetics inevitably ensue.

Therefore, in conclusion, the significant swelling of Al2O3 alumina at high temperatures is attributable to the unfaulting of interstitial dislocation loops and the subsequent formation of dislocation networks, which serve as efficient sinks for the absorption of intersti­tial atoms. This leaves behind a supersaturation of lattice vacancies, that is, an excess of unpaired vacan­cies in the bulk of the Al2O3.

In irradiated MgAl2O4, only high-energy faulted loops are available as sinks for interstitials. Therefore, in this case, interstitial-vacancy (i-v) recombination is the dominant mechanism for defect accommoda­tion, and negligible swelling results.