I have read here that the best way is to keep it simple, be short and ask for meetings, which is a bit generic way to ask for a job. I agree more with the discussion posted here by a student in a similar situation as mine (I still haven't started emailing people but I've already have a long unfinished list).
Speaking with some PhDs and postodcs I know, they told me that the best way is to know somebody in and maybe write them first in order to introduce you to the PI before you send your email. Another advantage of this is that you will know if there is a position available or if it's not a good idea to work there (the people who actually work there might have the best advice about that). Finally, they might also be "more reachable" than the PI who gets thousands of mails. On the other hand, I don't know the people who work where I want to go so it would feel as impersonal as this question feels to you.
Given the fact that I may have to write several (around 20) of these mails I want to have a good strategy. I know it's a debate that mostly depends on the area. My area of research is Neurobiology of Memory and Learning but I might also be interested to change a little bit, maybe to Addiction (Always within neurobiology).
I would also like to add that in order to write a detailed email about the PI's research history and future path one must be able to have the time for:
- Read the publications = A large amount of time
- Actually come out with an idea that meets with the PI's expectations AND research interest/approach/knowledge AND more important technique capacity
Of course, being a PhD student means that you have to suceed in both of them but not for every research we would be happy to do (if we were accepted) in advance.