The panel retaining a bias. LCDs are driven with AC waveforms that get switched around to try and work around this persistence thing (just read any old panel datasheet and there'll usually be a pin doing that.
Mono polarizers look different because of a retardation film, this is just a filter layer.
I have a vinegar'd 180 panel - the one I intend on using for testing my ideas on vignetting (that's currently 8000mi from me) that would like a word about that.
The charge retention is why the moisture invasion from a bad seal for the Hosiden active matrix mono LCD's makes more sense than the competing theory for tunnel vision. The other theory has no mechanism to explain how the charge retention occurs. For the moisture invasion theory it is easily explained that the moisture can create a parasitic capacitance which temporarily provides enough energy for image retention. Eventually the parasitic capacitance loses its stored charge and that is when the retained image is lost. That's why it isn't a true "burn in", it's a pseudo "burn-in" because the retained image will dissipate if disconnected from any power source for several days. I have been able to demonstrate this when a Hosiden LCD from a Powerbook 540 was completely stripped all the way to the down to all its various individual pieces of polarizers and diffusers along with the glass sandwich that contains the liquid crystals and TFT. I was able to show the retained image in the tunnel vision region by replacing the polarizer on the glass sandwich, showing that the retained image is inside the glass sandwich when it is attached to nothing, indicating that it is within the liquid crystal layer itself. The only way that would be possible is if the source of the power for the liquid crystals was within the glass sandwich. Parasitic capacitance in the liquid crystal layer makes the most sense for how this is possible. The best answer for parasitic capacitance in the liquid crystal layer is moisture intrusion.
Anyone can test this for themselves. Take a Hosiden active matrix mono LCD that hasn't been baked and has a retained image after being powered down, strip it down completely to all the individual components of the screen. Put the polarizer back over the glass sandwich that isn't attached to anything and you will see the retained image. Give it a few days at the most, and you will see when placing the polarizer over the glass sandwich now that the retained image is lost, because the parasitic capacitance has dissipated. The competing non-moisture intrusion theory for tunnel vision can't explain this behavior with their model of how tunnel vision occurs.
As far as baking a screen in a vacuum, as far as I know, no one has been able to do this due to a lack of proper equipment.