The spectrograph array was exposed to continuum light such that the spectrum was nearly saturated at points in a 1 second exposure (35000 DN = ~100,000 electrons).
The light was turned off and the entrance aperture blocked with foil. About 1 minute afterward a series of 5 minute (Fowler12) exposures began. The plot below shows the level of persistence in the brightest part of the spectrum (where the counts were 35,000 in the 1s exposure) over the course of 10 five minute exposures. Even after 50 minutes the persistence level (0.12 e-/s) is at or above the dark current level. The estimated gain of the system is 3.2 e-/DN.
If you plan to do sensitive spectroscopy keep bright sources (especially calibrators) off of the array for at least an hour beforehand.
Note that thermal K-band emission still appears in all of the images below in the upper left. The persistence signal is that in the order below the K-band oder at the top.
Persistence in the first 5 minutes
Persistence after 50 minutes
Still to be done - evaluate persistence for different illumination levels to determine a maximum flux (DN/s) to be avoided.