I'm astonished there's not an equivalent for this literally everywhere. Even if some articles seem not terribly important at the moment, as progress marches on many could become relevant again and losing access to those is simply inexcusable.
Work funded by the NIH needs to end up in PubMed Central within a year of publication. The NSF and DoE have similar policies. Unclassified DoD-funded work also needs to end up in the Defense Technical Information Center. All of this flows from a 2013 memo by John Holdren/OST entitled "Increasing Access to Results of Federally Funded Science", and, as far as I know, it hasn't been overturned. While this doesn't formally cover everything, it comes pretty close and many journals now handle this automatically.
The memo is here: https://obamawhitehouse.archives.gov/blog/2016/02/22/increas...
Canada has a similar policy for Tri-Council funded research (here: http://www.science.gc.ca/eic/site/063.nsf/eng/h_F6765465.htm...) but does not require specific repositories.