The symptoms are very strange. My bet is no one else has seen something like this before. I sure haven't seen it in a sensor. The symptoms are confusing in that you have only seen it is the Dark Frames rather than in all frames. How closely have you really looked? Do you have PixInsight? Maybe a close look at the light frame using pixel by pixel samples would show a slight change in the values at the discontinuity.
You say this happened suddenly and had never seen it before. Were there any changes to your normal procedures around that same time? Did you notice anything odd about the hardware around the time of the appearance of the artifact?
Is there a time component to the severity of the defect? In other words, if you were to power up the camera and take a dark, then power down the camera for a minute, power it on again and wait for one or two hours and then take a second Dark Frame, are there any differences between the frames? You might want to try running that test and uploading the two Dark Frames for others to examine. For that matter, if you can upload one Dark Frame, one Flat Frame, and one Light Frame, I would take a look at them for you.
I have only a wild, wild, wild guess at what might cause an artifact like what you show. Many years ago (about 40), I was tasked in my job to find the root cause for a prevalent memory DRAM failure. The problem was such that tests run shortly after power on would always pass. However, if the system were allowed to remain powered on for 24 hours, some memory storage locations would fail to be able to reliably store a bit of information.
Working with the silicon foundry making the DRAM chips, we discovered the cause to be a domain boundary problem in the silicon the chip is built upon. Charge migration in the bulk substrate caused charge to pile up at the boundary between two crystalline domains within the silicon. This bulk charge build-up in turn caused the transistors nearer the surface of the chip to change characteristics -- they did not operate as they should. If power to the chip was removed, the bulk substrate was able to drain the stacked up charges and when powered back on, the chip would work perfectly again for a period of time before once again becoming unreliable and failing.
As I said, this is a really wild guess as what might cause what you are seeing in the sensor. It must be something new since if the bulk silicon wafer the sensor was built upon had the above problem, it would have always shown it unless maybe you have changed you normal procedures and now run the camera for very long periods of time before doing your Dark Frames. However, there might also be a very remote chance that the sensor has sustained damage to the substrate. Theoretically, this could be caused by extreme thermal shock although that very rarely happens because of the thermal mass of the chip and especially the carrier it mounted on. If a small crack has occurred in the substrate then it is possible that charge build-up at the resulting boundary might cause an effect such as you have shown. If the symptom turns out to be a (partial) crack in the substrate, then you might expect the artifact to get worse as time goes on and the crack propagates to the surface where it will affect the electronic structure of the sensor itself.
This is all very thin on evidence but it might be possible to test for a time-related symptom that could help pin down whether this is a sensor failure.
Edited by jdupton, 23 August 2019 - 11:37 AM.