Scientists write them for free, to embellish their resumes (often their grad students do the work) - the publisher sells a thousand copies, mostly to libraries across the world.
Rinse and repeat.
The problem with that idea is that there would be very few copies, concentrated in the urban areas that are likely to get nuked. A book like one of those is also likely to be extremely specialized, so it would be unlikely to be used if that specialty ceases to be of practical values for a generation or more. Even physical books need care to be preserved for long periods, and wouldn't survive in an abandoned library with a leaky roof and flooded basement.
I agree, but "in in near-apocalyptic circumstances" they'll only preserve the stuff they value or find useful. I don't think redundant, low-print volume academic specialist works will fall into those categories. Unless there's enough surplus that the university continues to function as such, those books (along with dissertations, most academic papers and archives) will rot on the shelves.
In the aftermath of a full-on nuclear war, I'd expect the English-speaking word would experience huge, unrecoverable losses in areas beyond the undergrad level. Advanced work that's famous and highly-cited enough to be in a South American or African university probably would be an exception, though.
That's a separate problem. You pay to submit the paper and then people pay to read it.
I really wish we could expand human knowledge without gatekeepers milking both sides