From Prokaryotes to Eukaryotes--Development of the Nucleus

The nuclear membrane breaks down during mitosis.Thus there is no need to actively cross it while intact, since the nuclear DNA is exposed (and at its most vulnerable for insertion events) during mitosis. I'm not sure what the prevailing theory is here, but it is likely incorporated during DNA replication, essentially as a mistake in a recombination or DNA repair process. (I'll poke around on the net and see if I come up with anything more specific to mtDNA).Keep in mind, this isn't something that happens "regularly", so it may not have a standard way of occurring; rather it is the result of occasional genetic accidents.

You seem to be on the right track, that an error during replication (false recombination or strand break, perhaps) opens up the opportunity for an insertion to take place.Also think about secondary and tertiary DNA structure (like "hairpin loops") - DNA isn't always a nice double-stranded helix, especially when bizarre things are going on. During an insertion event, DNA at the site may give up its simple double-strand pairing for a more complex folded structure to compensate for basepair mismatches. The folded structure only needs to persist until the next round of DNA replication, when each half of the strand serves as a template to produce a complementary strand.If you are still interested in more details, you might try searching about for information on "transposons" or "mobile genetic elements", which are genetic sequences that are especially good at inserting themselves in the genome, and sometimes multiplying once they are there. A decent amount of research has been done trying to characterize the mechanisms these elements use during insertion - I believe use of secondary structures may be involved.