If your heater bed isn't in contact with the glass plate, then you are depending on heat transfer across a layer of air which is a great insulator. Glass is a poor conductor of heat, too (0.96 W/m-K; air is 0.02, copper is 400 W/m-K), so heating one region of the glass plate does little to heat the adjacent regions. So what you really need is contact between PCB and glass all around the glass plate. Remember, that thermistor tells you what the temperature of the PCB is in that specific area of the PCB; the solid copper layer should do a pretty good job of distributing the heat, but you need to get that heat from the PCB to the glass evenly as well.

When I assembled my 10" i3V, I noticed some warp in the heater PCB and corrected it by gently flexing it. I think that ideally, you want the PCB to be convex up so that the glass plate sits on a very slight hump in the middle, and then you use the clips to pull the edges into contact as well. The middle hump (or ridge, in my case) keeps some pressure against the middle to maintain contact (not too much, you'll flex the glass). I find that clips on all four sides are usually needed to get good adhesion near the edges, too. The LAST thing you want is for ANYTHING to be between the glass plate and the PCB. You might try a very small amount of heatsink grease to fill very small gaps, but that's a poor substitute for direct contact. It's probably more useful as a visual to see where the glass isn't touching. Where you have good contact, the grease should be squished thin enough to see the PCB through it. Where all you see is grease… need better contact.

The nice thing about this PCB / glass bed scheme is that you can replace the glass when it breaks. Otherwise, I'd be tempted to simply glue nichrome directly to the glass.

--Rob