Is it necessary to report a researcher if you are sure of academic misconduct?

Maybe

It is possible your institution will require you to report fraud at the institution if you're aware of it. Here's an example from the University of Manchester (clause 5.2).

All employees and students, including those holding honorary contracts, are required to report, and all other individuals working at the University have a responsibility to report, to the University any concerns about the conduct of research (as defined in section 4), whether this has been witnessed or for which there are reasonable grounds for suspicion. Failure by a member of staff or student to report potential/suspected research misconduct may constitute concealment of research misconduct as defined by 4.2 (f) of this Code of Practice.

But if you're just a reader with no relation to the authors in any way, there's no requirement you report them. You could still do it of course, but there's no requirement.

I'd not contact the author, because there is a very real risk of retaliation. Contact the author's institution or the journal (or maybe Retraction Watch) instead.

“ What I noticed first was that all figure panels were messy, figure legends were nonsense, and the text was very weird”

First, poor production is the problem of the journal, not yours.

Beware as some near-duplication may be acceptable. Some IEEE journal for instance accept “expanded versions of conference publications” which may themselves be published as IEEE proceedings so what constitute an “expanded version” is not clear-cut.

Image duplication can be the basis for retraction but usually this is because the image is used to support fraudulent conclusions. In itself, the same data (and thus the same graphs) could be used to analyze different models and in such cases duplication can be acceptable. The same holds for figures: the same setup could be used twice, or a nearly identical setup used with minor differences (say: changing the input or the detectors or whatever) so that using twice a figure illustrating the same setup may be legitimate.

Be careful before concluding or even alleging academic misconduct. Errors happen in good faith. If you have serious concerns, you should contact the journal and let the journal investigate the situation: a journal could be in legal trouble if they unknowingly published material copyrighted by someone else so it is in the interest of the journal to make sure such allegations are confirmed or voided

Yes. Contact the editors of the journals

Your comment, "I checked again and found that one figure was used at least 4 times in different papers, all the times it was supposed to depict different materials" is the critical piece of information that makes this unambiguous. Contact the editors of the journals those specific figures appeared in, and provide them with the list of papers, figure numbers in each, and specifically mention that all those figures are supposed to be different materials. Mixing up figures once could be coincidence. Four times is almost definitely deliberate fraud.

Do not contact the author. They're the villain of the piece, and you don't want to give them a chance to retaliate (which similar frauds have done in the past). Go through the editor and preserve your anonymity.

I'd only really be concerned about this if what you have identified is fraud aimed at making the world believe something incorrect about questions in the field. In this case, by that I mean have they used the same figure twice, but claimed it is showing different things. As you say, if they've said in one paper "Figure 1 is SEM of foo" and in another paper they've said "Figure 2 is an SEM of bar" and in a third paper they've said "Figure 2 is a TEM of a foo", all three figures are the same image, then yes, this is fraud.

Fraud should ideally be reported. I'd report to the editors of the journals concerned. You could also try reporting to the standards officer at the researcher's institution.

If this is only happening in poor quality journals, I'd guess nothing will happen, as the journals probably don't care. If this is affecting the papers in one of the higher quality journals, they MAY do something (publishing fraudulent papers in one journal will not get papers in another journal retracted).

All the other issues you mention see indicators of a poor quality scientist, but not necessarily a fraudster.

The only benefit I can see to contacting the researcher directly is if somebody has effectively stolen his identify as part of a citation cartel.

Are you absolutely sure of academic misconduct? It is possible to publish poor quality research, but this does not necessarily mean data fabrication, if that is your concern. There are likely a lot of QRPs here, but I think the evidence should be stronger to make a definitive accusation. For instance, looking at how they report their methodology, effect sizes etc. if it is poor/absent, then it could be hard to say. However, if you are able to deduce from how they report their methods inaccuracies or inconsistencies, you may have a case.

What you want to do about it is up to you. If you think it is harmful to your field or valuable research, then it may be worth considering reporting. However, it would be unusual for that many publications to be submit in high impact journals if there were no formal review process. Unless somehow they are a mastermind at data fabrication, in which case it would probably be a lot simpler to report real data.