It can indeed be done for free, and in fact there are some totally free journals that are indexed (actually I am using "indexed" as shorthand for clarity/simpli

It can indeed be done for free, and in fact there are some totally free journals that are indexed (actually I am using "indexed" as shorthand for clarity/simplicity, because it's a bit more complex than that: it needs to be indexed on a certain database -proprietary, by the way- and ranked with a high enough impact factor - a crude metric based on citations per paper). The problem is that they are still a minority, and often if you are working on subject X, it turns out that no indexed journal on that subject is free. Or maybe there is one, but your paper gets rejected from it so you then need to resubmit to a non-free one. And why do publishers like Elsevier hold most of the indexed journals? Pure inertia. We keep sending papers there because we need to for reasons like the ones explained above, then our papers get citations, so those journals keep their impact factor high. And it is extremely difficult for new entrants to break into the rankings, because sending a paper to a new journal that is not indexed yet has a huge opportunity cost (months of work that you aren't going to get recognition out of for tenure, promotion, grants, etc.) so unless they start with really great momentum (support from a powerful scientific society, an open letter by big shots in the field, etc.) they never get the needed papers and citations to climb the rankings. So it's not that publishers do a better job so their journals have higher impact factors, it's that they have a captive market.