Perhaps they are taking oxygen or carbon from ambient, you can put them in a desecator with vacuum or in a chamber with an atmosphere controlled of nitrogen or argon.
Check them by EDS if there are impurities of carbon
Check them by X-ray to see if there are changes in structure
Also UV-VIS and Raman can give you an idea of what is happening on your samples.
Delayed cracking is typically attributed to hydrogen embrittlement. Not sure if hydrogen was introduced at any stage of sputtering or was present in the targets. Another source for cracking is residual stress or thin film growth stress. Look at the mismatch of the coefficient of thermal expansion of your film and glass substrate.
In sputtering depending on RF power one has to see the target backside cooling attachment, type of target (oxide or metal) is also play significant role to induce thermal energy.. I think operating power is to tuned according to target type....