I promised an answer, but I already made half my points in comments. I suppose I will make one more one. Most people have given justification for 'monogamy' but not life time obligatory monogamy like you want, so presume the answers above all apply to encourage serial monogamy, what will force a life time pair bonding?
Paragraphs in parenthesis go into more detail explaining likely confusions about evolution in general, they better explain why an idea makes sense evolutionary if confused, but can be skipped for those familiar with evolution.
Honestly, as I already implied it's pretty hard to justify. I do have an answer, but unfortunately I'm sticking to evolutionary principles, and the best way I can justify mandatory monogamy starts from a rather unfortunate start so umm...My answer is rape...lots of rape...cough Hear me out first there's a reason here...
to be more accurate I'm referring to rape in the distant evolutionary past, particularly due to the resulting evolutionary arms race. Males try to rape females, females want mate choice so they develop means to prevent undesired pregnancy even if raped. This sort of arms race happens quite often, though less often in mammals.
A side effect of this sort of arms race can lead to females being less fertile, they are so good at preventing unwanted pregnancy they start preventing wanted pregnancy as well because their defensive mechanisms are too strong. Again, this is not unheard of and has occurred in many species.
( why this happens. In an 'ideal' evolution a female would evolve a way to prevent pregnancy from rape without preventing pregnancy from desired matings. However, evolution is not guided, or quite so 'smart'. It is very hard to evolve such nuanced defenses. Instead it's likely that if rape is harmful enough to female they will ramp up defenses against rape, as an undesired, but better-then-the-alternative side effect successful matings from desired mates may also drop by a smaller amount. After this happens an independent evolutionary phase may work to help increase odds of successful matings with desired partners. Many species have these mechanisms, most obvious being numerous species that will abort pregnancy if exposed to a foreign male due to the odds of a foreign male killing their child after born. This can cause undesirable abortions at times when the foreign male was not a threat of killing the resulting child, but it usually saves resources to abort and so it was evolutionary favored even if it sometimes harmed mating success. After such defensive mechanisms exist later mechanisms can evolve to help better distinguish which mating's are desired if defenses are hindering successful matings too much. Which brings us to the next point...)
So now the evolutionary problem to solve is how to allow the mates you want to mate with to have success when you have such strong defensive mechanisms. The answer is to have means to help accept a mate that your body recognizes. Pair bonding can be part of a process of creating not just a psychological but also physical connection to a mate which will actually help insure female anti-rape defensive mechanisms will recognize her mate's sperm and not attack it.
(To be more accurate this isn't a simple chronological progression from A to B. Instead this would be a cyclic process, where increases in A or B encouraged further increases in the other. As The first anti-rape defenses likely only prevented a small number of matings, say 10% of undesired matings and 4% of desired matings. The original difficulty with desired matings put pressure on evolving the first limited mate-selection methods. As mate selection methods grew better it lowered the 'cost' of anti-rape defenses, as they didn't hinder desired matings as thoroughly, and so these rape defenses ramped up further, encouraging mate-selection to further evolve etc etc. As those defenses further increased the need to better identify desired matings increased and so on, each time one trait grew more refined it allowed, or encouraged, the other trait until eventually both were fully formed.)
Mate selection mechanisms are not unheard of. Kissing in humans is not just cultural, it's evolutionary practice shared with many primates. There are now many evolutionary advantages to it (once something exists in a species they usually find multiple ways to exploit it), but the original cause of kissing was likely to help women be able to carry a mates child to term. In this case this was due to a very common STD that was dangerous if caught while pregnant or when conceiving, but not before. The best approach for primates to deal with it was was to intentionally catch it if you mate has it, via kissing and swapping saliva, when it was harmless, so that by the time your ready to mate you already have CMV and thus don't risk catching it when mating which could result in severe birth defects to the resulting child. The point being kissing, a romantic gesture now, started as nothing more then a physical act to help ensure a success when mating with a chosen mate.
I would do the same with pair bonding, it is a way of helping the body recognize the mate so that mating success will exist later. I would change one thing, remove pheromones (which radiate out from someone so you couldn't have pheromones work on one person and not anyone else near by anyways) and make the bonding a more physical act, something that allows exchange of DNA so the female's body & immune system can start to recognize the mate. The obvious answer would be to use the above example, kissing. Kissing is the start of getting DNA from saliva to start recognizing your mate. It's not love at first site, it's love at first kiss.
(note, it would have to be more then just kiss or a male can force a female to kiss him. However, if it requires a long continues length of time to build up the bond a rapist isn't going to hold a female captive for months to develop sufficient compatibility to mate)
The more mates are close together, and exchange err 'body fluids' through kisses or otherwise, the more the female immune system recognizes the male and the better the odds of mating success. Thus life-time monogamy may be preferable because the odds of successful mating increases the longer the pair is together.
From an evolutionary perspective a female may benefit from a system that causes a mate to spend extensive time with her before mating is guaranteed, not just to prevent rape but also to have more time to 'judge' the fitness of the male, to gain benefits from courtship (which likely involved male offering 'free' resources to the female to earn favor, and to encourage the mate to stick around to raise the child since they now have a better chance of successful mating with the female in the future. This is important because it would give evolutionary advantages for women to maintain traits that lead to pair-bonding even after the threat of rape was mostly mitigated.
I would couple this with other mechanisms to prevent cheating, particularly to prevent females from consensual matings with males. as I mentioned cheating is quite common in all monogamous species and would discourage such strong pair bonding in most species. Have males have a method of reliably recognizing their children with a very high degree of accuracy. If a female births a child that is not biologically the males he will usually kill the child immediately in a rage. To make this work you would need to imply certain specific markings are carried exclusively on the male chromosome so that the male alone controls the marking and can recognize his marking with others (there is more to this, ask another question if you really want a full description for how a male can be confident he was not cuckold). This removes the advantage for the female of cheating on the male, and male would have little advantage of cheating since the female will rarely conceive from a mating with someone they haven't bonded with first.
The strong ability to detect paternity could have evolved back when rape was common due to the increased odds of cuckolding due to rape for males, and would have further driven females to excessive biological mechanisms to defend against conception during rape due to the gaurentee that the child conceived will be killed at birth by her partner making conception from rape an entirely expensive waste with no evolutionary success.
Eventually the rape-tendency will go away as it proves to not result in children, due to female defensive strategies, but the pair bonding will stick around for awhile longer.
This would also slightly encourage what you wanted earlier, about a mate not trying to mate again after their mate passes. I stick with the statement that this would not normally happen. However, if the mates are older when one passes away the remaining mate may find it not worth trying to find a new mate because it would take too long to build the bond to the point where a mating was likely to be successful, and no opposite sex would spend that long trying to build up a relationship with an older mate that will die sooner leaving them in the situation of old maid unable to find a mate. Thus if your mate passes after a certain age the remaining mate may find it adventitious to instead focus on supporting his current children and family to help them achieve better reproductive success because no mate will choose to start bonding with them so it's not worth wasting energy trying to court a mate. Younger mates will likely still court new mates if their first bond dies young. Then again, if the bond builds over time they may not have the same emotional bond yet...
Keep in mind the above only works if a mate that gives up on finding another mate is also driven to help it's family members find reproductive success, not just giving up and sitting around doing nothing with their life. That likely means providing grown children and siblings with food and resource etc so they can focus on raising their young.
*note, this implies a larger sexual dimorphism between males and females, with females being smaller, at least back in the more past, which is why females had to defend themselves via biological methods instead of claws and teeth. If you really want the two sexes to be close to our level of dimorphism there are some tricks you can use to justify dimorphism decreasing during the time it took for pair bonding to evolve, due to unrelated evolutionary pressures...
Now, why this is still a lie... There are a few places where I'm bending evolutionary principles a bit to 'fit' the desired result, since as I said I don't honestly think this would evolve. I think this is a best-fit option and you likely can get away with handwaving the cheats, but I want to at least mention them so you are aware. Here is a quick list of my cheating:
Rape-mitigation strategies usually involve physical changes, not immune system
usually counter-rape evolution involve more complex vaginal tract that make the act of mating harder for rapists, or some way to stop the sperm of the rapist from entering the main reproductive tract. This is easier to evolve then some sort of immune system response to stop rape as in my suggestion.
To counter this I would combine rape-migration strategy with disease migration. Some strong disease forced heavily ramped up immune responses in the species about the same time that rape-prevention mechanisms were evolving. The 'accidental' result of the ramped up immune system attacking sperm had the side effect of working as a rape-migration strategy which it got selected for.
To make this work better I'd suggest an STD (or probably a collection of rapidly evolving/diverging STDs) being the cause of the ramped up immune system, to better explain why the ramped up immune system would attack sperm in particular. To make this work better you would need to explain why rape was still worthwhile to males if there was a non-trivial risk of STDs. Best explanation I can think of is a particular STD could be transferred from male to female via ejaculation but either did not migrate from infected female to male or, more likely, did not harm the male as much. For instance perhaps the virus only harmed a female's later reproductive success by making future implantation harder (but not ones resulting form the rape since the virus needs more time to spread before it has adverse effects). This would put the virus as a negative to the female but not the male, causing males to feel free to rape but females to develop strong STD immune system, that then also got specialized into an anti-rape defense.
Evolutionary pressures aren't enough to push such a strong extreme
This is my biggest cheat. All of this process could start as described, but probably not end like this. The problem is at some point women would become 'good enough' at preventing rapes that rape wasn't cost efficient, and at that point they would stop happening. This point would likely be before pair-bonding became 100% mandatory for successful matings. There are costs to this system as I defined it, and at some point rapes would be uncommon enough that they didn't warrant further opportunity costs, or simply didn't provide sufficient evolutionary pressure to get out of an evolutionary 'valley'.
There are some things you can do to mitigate this risk, though I like one most of all. I suggest adding a competing evolutionary cousin living close to your species, one that is still very rape happy. The cousin species would attempt matings with females of your chosen species. These matings could lead to pregnancy, but rarely fertile children.
This helps to ensure that females always need anti-rape defenses, because even after males of their own species have evolutionary pressure to stop raping the cousin species, who's females do not have sufficient anti-rape protections, may still maintain rape as a successful mating strategy and thus the males of the cousin species may continue to rape females of your target species.
The males of the cousin species would not be as tempted to mate with females of your species, Kinophillia in general would make the females less attractive and they would not be gaining any major evolutionary advantage from it. However, they would not be suffering any significant cost for the attempted rape either, meaning there is limited evolutionary pressure towards the cousin species males evolving an instinct to avoid raping the females. Put another way raping of females (the ones of their species) works well enough for there to be strong incentive to evolve a 'attempt rape of any available female' instinct, and there may not be enough of a cost to failed rape attempts for males to evolve a more nuanced understanding of which females are worth mating and which are not.
The males may also gain some secondary evolutionary advantage from attempted rapes of females from your target species. For instance perhaps it serves as practice for attempted rapes of females of their own species, with the learning experience being enough to incentivize attempted matings continuing.
This isn't a perfect solution, since it's likely that females of your target species would evolve simply evolve a means to prevent pregnancy from ever occuring from matings with the cousin species (Since there is a difference in DNA between the two species any number of mutations may prevent hybrid children from being viable but not affect success of matings with males of the same species). Still, this could place an extra layer of pressure on the females at the same time attempted rapes from males of the same species, even if decreasing in frequency, was adding it's own pressure to help further push towards stronger immune response.
For this idea to work the cousin species would have to have once been reproductively isolated, long enough for them to split off as a separate species, but that they now are less isolated. They likely were isolated for a relatively short time, thus the reason they still are similar enough to keep kinophillia low. This suggests either one species migrated away from the other, evolved into a new species, and then migration brought the two back into contact. Alternatively something 'temporarily' isolated the two species but limiting their ability to travel. For example after an ice age a river fed by glacier ice water grew too fast to be able to cross, separating the species, then after all the glaciers melted the river flow dropped down enough to make migration across the river easy again.
A separate solution from the cousin-species one is to use my CMV example above. Make mate-selection partially about getting immune systems in sync to counter virus or disease. If you do this the original mate-selection process may not be just encouraged by rape by by immune system merging. I'd use both options honestly, you need all the excuses you can to encourage such a strong evolution.
There is no reason for both males and females to have anti-rape techniques
If males could successful identify cuckolding and kill the child then there would be no reason for rape as the child would not survive. It doesn't make sense for both female and males to have such well evolved anti-rape techniques since either one alone can make rape evolutionary disadvantageous.
This is partially fixed by saying males are not always able to identify cuckoldry, or sometimes are less confident of when they identify it. while they sometimes kill children they are certain are not theirs, they also often abandon the female with the child, when they strongly suspect, but can not confirm, cuckoldry occured. The combination of possible infanticide with abandonment combine to put a heavy cost on females for having a child conceived from rape while giving enough of a chance for the child to survive to adulthood that rape may still be evolutionary advantageous to males.