Originally Posted by deepor
I don't think I can find those experiments with Intel's TIM. I think it was somewhere here on this forum. The way I remember it, the person simply tried to put the IHS back on after cleaning off the glue, but not cleaning off the TIM.
You might be misunderstanding what people here think when they talk about a gap between die and IHS. It seems it's known that when you look at the parts when they are completely clean and without TIM and without glue, the gap you will see is between the IHS edges and the PCB while there's no gap at all between die and the middle of the IHS. At least this was the case a long time ago with 3570K and 3770K. There's was this post here four years ago which shows this in a neat .gif:
It makes it obvious how the parts are designed. In the design, it seems there's not supposed to be any gap between die and IHS. It's just that, this is not what happens in practice. After a CPU goes through manufacturing, it seems to usually end up with a gap there.
That's perhaps where the misunderstanding here happens? When people talk about there being a gap between die and IHS, they mean there's a gap in practice. They aren't talking about the theoretical design.
There are tools for measuring gaps, example;
One can also use a piece of paper in this case. If the wafer is clean like this;
And the IHS is equally as clean where it counts most (no picture, use imagination), then it stands to reason that there can be no gap
between IHS and die if when the IHS is placed over the die, there is a gap
all around between IHS and wafer. The IHS cannot levitate, it has to be resting on the die. The base of the IHS has to be sanded down in the unlikely event that there is no gap between IHS and wafer.
This is what one should ascertain before applying the new TIM and sealant. Pressure has to be applied while the sealant is curing so that the sealant thickness between IHS and wafer, where it counts most, is no greater than the original gap
. Excess sealant will be forced out of the gap on both sides, while in its liquid state, long before the sealant starts to cure.
Now off to watch something on Netflix.