when a part is forged, it's like wadding up a piece of paper the paper is stronger because it's denser, but the structure inside that paper isn't as strong as it could be because the outside molecules are squished together and the inside is closer to its original dimensions. When a part is cast, the molecules stay equidistant in relationship to each other, thus its strength is the strength of the material. Heat treating (and cryo treating) attempts to regain some of the lost strength that is caused by the squishing (technical term
) the metal together like paper - what it does is bring the material up to almost-melting temps and allows the molecules to get back to equidistant spacing. That is stronger, and retains some of the plasticity of the steel. Cryo treatment is different because it work-hardens the material by freezing it - when you bring the steel to super-low temps, it causes the molecules to push against each other, thus work-hardening the material - they only put the material in the fluid for a set time to limit the depth of the material. Picture baking a cake - if you pull it out too soon, the inside is still moist, too late, the outside is crunch hard.
The advantage of work hardening the surface is the parts have less friction, thus less heat. Less heat means longer life and better dimensional stability. The disadvantage is if you do it wrong, the part becomes brittle - without lab analysis, I suspect that's what happened to your part. They cooled it too long, making the surface very brittle and losing all the plasticity of the forging/heat treatment process. It's not a rookie move, the science of cryo treating is a very exacting science, and you can't treat a part using analogy.... example, you treat a camshaft for 10 minutes - by experience you know that gives you the best hardness without losing the ability of the cam to flex under load.. thus, not snapping like a toothpick.
While I understand your frustration with the CI differential, when you said that your half-shafts have survived longer - you made me suspect that all is not well within your differential, thus making it more likely to break. Back in the early 60s, my dad used to drag race a 39 Chev coupe at Lions - the rear differentials they used were smaller than what a rwd corolla had... and they survived because the people who drove the cars knew their weakness....
please forgive the shortness of the response, there is a lot to metallurgy ... I've been playing with this stuff to make my 455 Buick stay together under 900 hp of boost... on a cast crank.