In this work they pit a nice shiny Exploris 240 system with capillary flow (1 microliter/minute, which I'd still consider nanoflow vs microflow (50uL/min) in a variety of different contexts across multiple sample loads and total experiment acquisition times.
It is largely what you're expecting --- peak shapes are better at higher flow rates, and lower flow rates do better with lower sample loads.
Where all the work this team did really shines is in where you can get close to matching the coverage at these high flow rates vs low ones and you find these points where it seems pretty darned smart to use the 50uL/min.
For example, for 15 minute LC runs, 2ug or 5ug sample load is pretty darned close at 50uL/min or 1uL/min. You could argue that was probably just too much sample for the little column used for the 1uL/min gradient, but still...we're talking about almost 2,000 protein IDs in 15 minutes! That's legit.