There is a nice calculator for this on the CAOS website
which I believe is incorporated in to Ken Harrison’s modified version of Christian Buil’s simspec, the spectrograph design spreadsheet. Plugging in the values for my setup (C11 at f10 and 3 arcsec seeing) it calculates 50% throughput at 23um, 70% throughput at 35um and 85% throughput at 50um slit width. I normally run a 35um slit.
Note that other variables such as guiding accuracy and position of the star on the slit can be as, if not more important though. The definitive test as to how much light you are using is to measure the flux of the spectrum of a star with and without the slit. (or in the narrow and wide parts of a photometric slit of you have one)