Stress-free calculations. We have redone the stress-free bias factor calculations of Xxxxx et al. [11] for the three dislocations, using the analytical elasticity theory considerations and the finite element method implying the optimized numerical procedures, thus improving the accuracy of the results. Also, we have redone the stress-free atomistic calculations of the BF previously executed by Xxxxx et al. [11] applying the extended relaxation procedure for the crystals as it was described in section 2.2. The new atomistic interaction energy maps for three dislocations with XXXx or vacancies are presented in the left columns of the Fig. 8 and Fig 9, respectively. One can see, that compared to the results in [11] the atomistic interaction energy landscape became more symmetrical (mirror symmetry along [111] with respect to the dislocation core position) for the edge dislocation with BV=½<111>. Nevertheless, the atomistic dislocation bias factor value (shown in Fig. 3) of 0.055 is very close to the value of 0.060 obtained in [11]. For the edge dislocation with BV=<100> compared to results in [11] the bias factor has dropped by 0.015 for all the temperatures considered and became negligibly small at 723 and 823 K. Interestingly, that unlike in [11] we have found that the analytically calculated bias factor of the edge dislocation with BV=<100> is lower than the one of the edge dislocation with BV=½<111>. Intuitively, the opposite would be expected (see discussion in [11]). The obtained result implies that both our atomistic and fully analytical calculations of bias factor contradict with the early assumption that the BF value should be proportional to the modulus of the BV [26]. Note that in the experiment by Xxxxx et al. [27], no clear correlation between the density of dislocations with a particular BV (<100> or ½<111>) and the measured swelling was established. Here, we also report the analytical and atomistic values of dislocation bias factor for the screw dislocation which was not presented yet by Xxxxx et al. [11]. Although the analytical predictions point to the very small positive bias factor value (below 0.01), the atomistic simulations show the negative bias factor of about -0.06 (i.e. preferential absorption of vacancies compared to interstitials). The physical model for void swelling in bcc metals which considers the negative bias factor for screw dislocations was discussed in [11]. Analytical edge <100> Analytical edge ½<111> Analytical screw ½<111> ρ =1013 m-2 d Atomisti...
