Como posso ajudar meus jogadores a mergulhar melhor no meu mundo de jogo?

Looks to me like you tried dealing with it in-game

And apparently it did not work. To me, there is nothing that you can improve as the DM, looks to me like you already tried being nice and patient about it ... and now it's about asking your players to do the same kind of effort for your sake.

It seems clear to me that their behaviour greatly impact your own fun as the GM of their group ... so you will probably end up querendo to leave this group if this aspect of your game does not change.

At this point, you should try OOC solutions

I would discuss with my players calmly and politely about how their lack of immersion breaks my fun as a GM. After having tried so many in-game solutions, it's time to have a serious discussion out of game. In the past, when I've had to deal with this situation, when players did not give a care about the World I put them in, having this discussion often helped. In my experience, that also requires patience, since your players won't change overnight.

Tell them you enjoy writing interesting stories, coming up with believable NPCs, etc. and when they end up just killing them right out constantly, it feels like preparing the game is a waste of your time.

Reassure them that this is not about controlling them and forcing them not to kill NPCs ever, but maybe their characters could stop being susceptible to the extreme and kill litterally everyone who breathes in their direction ?

You are entitled to having fun too. Tell them politely where your fun lies and what you honestly dislike ... which is easier said than done, but seems necessary.

Your players aren't interested enough in the world as you've prepared it, and are getting what they want de qualquer maneira.

Changing the world will be easier than changing your players. It's discouraging to hear, and I've been in your shoes, but at least for now your players sound uninterested in the setting and story that you've composed. Roleplaying and emotional investment require ongoing, active contributions from your players, and if they weren't doing that before then serving up more of the same is unlikely to produce anything but the same result.

From what's written in the question it simply seems that your game is set up to work with a fairly narrow set of play styles, and your players don't want to play in those bounds. The disconnect between your hopes and your players' behaviors comes in because the players don't want to play the game the as you've imagined it, but you still want them to succeed to see as much of the story you've prepared as they can.

The only solution I can think of is to adjust your story so that it can respond to the game your players clearly quer to be playing. If you can think of some plot hook that would attach the haughty, cutthroat player party to the story's events, and then think of how the party might behave while pursuing the plot, you can bridge the gap between your campaign and your players.

That will in turn make it possible to figure out what details will appeal to the players, making it easier to capture their attention and direct it. Figuring out porque this murderous band of people would be following your plotline will be more effective than constantly patching a plotline that necessarily assumes the PCs are different from what they are. If they are all on the margins of society, it's inherently challenging to make them care about the things from which they are disaffected.

It doesn't have to be guesswork on your part. You can ask your players what parts of the story they find interesting, what parts they're not responding to very much, and what elements they'd like to see that they think would make the story and/or setting more interesting to them.

Ultimately you can't force immersion or player engagement, no matter how enticing you make those.

So, tl;dr: Alter the story to be intriguing to the characters your players have placed in it, give bad behavior hard consequences obviously tied to that behavior (including campaign failure), or table this campaign until your players are in the mood to enjoy it.

A couple of more-specific observations I have that might bear on your situation:

1. There are no consequences for your players

You've described some in-world results of things your players have done, but it seems that you've been giving them story rather than penalty. Normally, that's great! But it's cutting against you here. For example, the demonic mage you described in the question dying and requiring an adventurous rescue doesn't sound like a punishment to me, it sounds like a fun quest which grew organically out of unpredictable gameplay events. It may be a departure from your story (and so a negative for you), but that kind of situation is the major reason that I (personally) like tabletop games: nada could happen, based on my decisions and luck!

And no matter how badly they bend your planned storyline, you are working to keep them involved in it. The ultimate effect of that behavior is that even playing as they are they can still complete the campaign and win the game. I think that it's valuable to keep a tabletop game world as responsive to your players as you can make it, but if you can't or won't stretch your story enough then failure (even irreversible failure) needs to be a possible outcome. Any other approach reduces immersion, because it makes it clear that the PCs really pode do whatever they want. Anything at all, except lose.

2. Immersion is built over time. It isn't something that just happens

I respect the time and effort you put into creating your setting and the adventure that takes place there. But it's important to recognize that immersion isn't about those things, it's about the interessado em e attachment to those things. Your game world is immersive to you because you've spent time and energy coming up with all of the ideas and then fitting them together-- if you look at some detail in your notes, doubtless you can immediately recognize its broader significance, its relevance to other events, what it means to different characters personally, and so on.

Your players don't have that. They have not invested enough time or effort engaging with the setting enough to care about the world or its inhabitants, and now that they've passed over 10 sessions' worth of material meant to help them do that it will be that much harder to accomplish with later material.

You can't fix that with more content, you need more noivado, which means more things that interest your players and fewer things that don't. If they only treat NPC companions as human shields, giving them more NPC companions to treat that way will only reinforce the idea that they are unimportant and expendable. A tweak, like requiring the NPC to arrive somewhere safely, can disrupt that pattern a bit, but you can't fazer players care about NPCs as people no matter how many there are or how detailed.

3. Consequences are often unclear to players

It's usually obvious to the DM how significant a player choice is, and in exactly what ways, but players are usually pretty clueless. It may be clear to your players in retrospect that they shouldn't have offed XYZ, because now he's the only one who would know what to do but is dead. Was it clear at the time that removing XYZ would be a problem down the line?

This was my biggest problem when I started DM-ing. I worked hard on building the setting, and so I knew very reliably what my players' plans were likely to produce. My players, however, often did not have enough information to make meaningful decisions. It was confusing for them and made the game less fun and less immersive.

4. Sandboxes can still have boundaries

I really like and value the possibilities that tabletop games bring to sandbox settings, but there's no reason you tem to maximize player freedom at all times. Even if the party could assault some NPC, maybe he's too well guarded to attack successfully. Or maybe he speaks to the party through an intermediary or unassailable barrier, so they can't actually attack him at all even as they interact. This is even more plausible if the party has an unsavory reputation (whether NPCs can prove that reputation true or not).

One of my best DM tools (in general, and especially against plot-destroying player creativity) is recognizing that while something in-game may be technically possible it can also be situationally impossível.

There's a mis-match in play styles here. You're trying to run a saga of some kind, and they're trying to play a sandbox. You aren't going to fix this quickly, but here are some possible methods.

The game world bites back

Stop bending game reality to avert a TPK. They aren't respecting the NPCs because they have no need to. They are doing murders and robberies with near-impunity, because you're trying to get on with "your" story and are trying to keep that on-track. Stop doing that and run their sandbox style of game, with its consequences.

Have the NPCs organise, arrest them, try them, and sentence them. Keep the trial reasonably fair, and have the various people they have wronged appear and give evidence. Remember, if you have Fale com os mortos available, killing people doesn't always silence them. Execute the ones who've deserved it. Imprison the rest who are guilty. This probably ends the party as a group.

Start a new party of characters, who may well be a bit better-behaved. If they aren't, repeat the process. If the players don't want to play this way, send them back to their computer games and find a new group. That's usually a lot easier as a GM than as a player.

I've never had to go quite this far with players, because I've always made it clear that the NPCs matter and aren't just there for killing, but it's always been my backup strategy. I have had local law enforcement execute one PC. The others watched, not because they couldn't have rescued him - they probably could, although they weren't quite sure - but because even though they were evil, they couldn't claim it was unfair.

Alternatively, give them another chance

Have someone hire them for a humanitarian mission. A new DM I played with recently did this - it was returning a child who'd been evacuated from a war zone to her family - and it's quite a decent way to give them a reason for behaving reasonably.

Give them opponents they can kill to keep them satisfied, but give the NPCs they talk to voices, names and personalidades. Have them be friendly, or hostile, or weird, but pessoas, who have lives of their own within the game-world, and aren't just there to play parts in your plot.

There is a question I often consider about characters with supernatural powers: If a significant percentage of the population can do this, then why aren't they ruling the world already? Most of these powers are well-known in the world, and it would be reasonable for anyone who wanted to protect their holdings to have countermeasures in play. The more common the power, the more reasonable it is that something exists to counter it.

One possible way forward for you, is to introduce an Authority that the Player Characters have no hope of challenging, and use it to guide them into a better game experience. I'm thinking of the film "The Usual Suspects", for example - a group of criminals all find out that they have offended the same legendary killer, and are tasked with repaying their dept or facing retribution.

In a fantasy role-playing game, you could achieve a similar effect by introducing a Patron of unfathomable power - e.g., an Archangel, a God, a Dragon, or just a king whose entire court are adventurers who retired at max level. Whoever you choose, make sure that they effectively counter the usual tactics the players employ. For instance, a King might be an unstoppable combat monster by virtue of being 10+ levels higher than the party, but his real power comes from a cabal of witches that give him absolute knowledge of the past, present, and likely future. His chamberlain is a high-level martial artist who can interrupt or disarm anyone who tries to make a move, without causing a scene or making an effort. His queen was the world's deadliest assassin, and still has needles and darts that can paralyze an ogre.

All this power is great - the players will have no choice but to comply. However, that's not enough to make the game work. We don't want to beat the characters in a fight, we want to convince them not to try.

The best way to establish the power of this Authority is simply to make it common knowledge. Each character has heard a rumor, or a secondhand account, or seen it themselves. "Your character knows this isn't a fight they can win directly - best to play along and wait for an opportunity."

The final step in using the Authority to improve your game is to use it constructively. Your goal isn't to punish or destroy the PCs. You want to point them towards adventure, glory, loot and legend. Like Keyser Soze in "The Usual Suspects", they have a job in mind, and it is one the players are uniquely suited for. Maybe a neighboring kingdom is becoming obnoxious - not worth the Authority going to war, but if the PCs were to go take out the Necromancer and recover his staff, the whole problem goes away.

If there are rules that the characters need to follow, simply make it clear that the Authority has an opinion. Killing in self-defense or in pursuit of the mission is acceptable, but unnecessary deaths make things harder for the witches. If the witches become annoyed, bad things will happen - curses, polymorph, people go missing in the night... The King shrugs. Not much he can do about witches.

The goal here is get the players moving in the right direction, not because they don't have a choice, but because you have presented them with a direction they want to pursue. Having enough power to enforce social norms is necessary, but using it should not be.

You need to do several things. You need to make them care more about NPCs. You need to punish murderhobos. You need to keep having fun.

1. Start using cards, flashy NPCs, and pop culture references. Your players seem to have a very short attention span. As such, you need to make it easier for them to remember key bits of worldbuilding. For key NPCs print off or draw a basic summary of who they are, what their capabilities are, and stuff. Print off and stick pictures of people they know from pop culture on for their faces mostly. Make very memorable characters. For key lore learnings you can do the same thing. If they learn a great secret, give them a card. Cards are much more memorable than words. Make sure the NPCs have distinct hair colors, personalities, traits, and all so they're memorable. Colourful, emotion filled and flashy is good. It also helps if you know your players, and make them similar to famous pop culture figures they like. I've done this to great effect when playing with children who remember nothing.

2. In the previous question I mentioned the idea of reputation. You need to make a targeted reputation. From the sound of it you've been twisting a lot to avoid tpks. That's fair. Don't twist to avoid death. Have gangs of fighters, angry law officials, armies personally angry with the murderhobo. When they meet the party have them explain that they have no issue with most of them, perhaps even offer a bribe to not fight. Divide and conquer. Evil players can get taken out, good players can keep playing. Let players face natural consequences. OOC, tell players this, that if others act insane you won't punish them for it, unless they choose to protect their murderous companions.

3. Keep it fun. The players like being insane murderhobos. Give them a chance. Give them missions that put them near weak, poorly defended settlements, isolated by monsters. There, away from the eye of the law and your plot, they can murderhobo with minimal consequences. Make it clear to them that murderhoboing in the big city is much more risky than in an isolated settlement.