units: 10^-6 m/(m K))

ABS:            73.8
Plate Glass:  9.0
Pyrex:          4.0

PLA is oddly missing from the internet(at least my brief attempts), but i found someone's MSDS that listed their "nylon grade" PLA at 9.0

Given a 150mm build plate and a 150mm print:  This means that given a cool temperature of about 50C (when i get impatient and start poking at the prints) and hot of 100C (my build platform temp of choice) that means that while cooling:

ABS will shrink 0.5535mm
plate glass will shrink 0.0196mm
borosilicate will shrink 0.0087mm

<values calculated with tool at above link)
It seems that although borosilicate changes at less than half the rate of plate glass,   The percentage difference between ABS and either make the difference seem moot.

Assuming we calculate stress at extrusion temperature (195c):

ABS will shrink 1.0515mm over 150mm extremely quickly when it hits the build platform (195 to 100c).

So the math says that for ABS (assuming I'm understanding it, which is rarely) plate vs boro differences are probably moot once considering other factors.

Now assuming that PLA value i pulled randomly is vaguely accurate.  Plate and PLA shrinkage basically identical.  Boro on the other hand shrinks more than 50%.   Does that translate to more self separation than ABS?   

Also:
Kapton (which is polyamide tape ..i think?)  rates at 110 ..so it moves more than ABS.   

I cant find PVA (which is the basis for pretty much all the hairspray, glue stick, elmers based washes)

PET (which according to amazon sellers is the greatest newest thing for 3d printing) runs at 59.4

So summary, I'm not surprised the kapton tape adhesive fatigues as it is moving almost as much as the prints do.

So now the new question is:

What is the optimum difference in thermal expansion that will ensure sticking while at the same time allowing a high probability of self seperation?

The funny thing is my PLA DOES auto-detach from plate glass. I'm guessing the opposite is happening.  PLA barely shrinks, while plate glass does. 

I need to do some math to see what kind of shrinkage we are talking about here.  Definitely sounds like it would be worth ~$10 for a test boro plate.

The other "magic" thing about it, and the initial popularity of it for Reprap, was a demonstration made by Adrian Bowyer of printing PLA on clean plain borosilicate. In my limited trials it's a little hit or miss, both by which specific glass I use (I have a few and swap them out) and how squeaky-clean it is before printing.

For my money with PLA I'd rather just use the glue stick then, too. But the expansion aspect is probably why the PLA does not just pop loose with cooling.

This apparently translates to 3d printing is that when the plate cools down it stays the same size.  when ABS cools it shrinks significantly.  Therefore the prints should self separate once cool.  This is something I've never seen with my thin frame glass.

that MIGHT be worth the money. Separating ABS from kapton is one of my least favorite chores.... assuming this is also not a myth.

(aside:  Why do my corners always seem to come up early or be permabonded to the kapton )

I have been using mirror tiles cut to bed size (cut myself using a cheap glass cutter) and they work fine, no cracks or breaking over a years use.

box of six 12" x12" plain edge mirror tiles: $12-15
cheap glass cutter: $2-3
4 printers with mirror tile beds, and heavy usage with no breaking: Priceless.

I keep seeing people say you must use borosilicate glass (pyrex,etc) for your heated bed.  I understand it is for safety aspect, but having a bed crack during a print doesn't seem to be overly dangerous to the print or the operator. 

When covered in tape of some kind, the glass wouldn't go too far if it cracked.  Even if  you were printing on plain glass, the print itself should hold the pieces together after a few layers.

Are there other reasons you should use one glass over another i am missing?