That would work, but it seems a bit "round-about". Consider these alternatives:

1. Align to an MNI template directly instead. There is no manual stereotaxic procedure available to go directly to MNI space, so you would only use the @auto_tlrc or the auto_warp.py script. Align each of the datasets to a selected template and then reapply the same transformation with @auto_tlrc for the affine transformation or 3dNwarpApply with the nonlinear transformation.

2. Use the electrode locations you've marked as tags. Mark the locations in each of the subjects in order with the Tagset plugin. Align all these together with 3dTagalign to a selected subject in native space, a selected subject in MNI space, an MNI template with the same tags selected. The alignment is based on the tag locations.

There are lots of ways to go between Talairach and MNI spaces. The method you show is one of several reasonable choices, but it will introduce more interpolation, and the transformation is an approximation. I think you would be better off with one of the other two choices.