Your chiller system probably has a limited capacity of water delivery, so at some point an increase of one partial flow can only be achieved by cannibalizing another one.
Jürgen Weippert Thank you so much for your answer. For my PVD system, water delivery is provided by a water bottle under required pressure. We have very same set up for the other oxide deposition system and historically we had never issued with water delivered to the system. But I will check if we can verify this idea. Thank you. Do you believe if there is any issue with target surface poisonings?
I mean, target surface poisonings may happen, but that should not affect the cooling water supply.
If you have an identical system with identical water input pressure, check the water filter or the tubing between the pressure gauge and the MFC for rust and chalk; eventually you may have some "internal pressure reduction" that your sensors might be missing.
Jürgen Weippert Thank you again for your insight. The process start nicely as in the figure for the different target but after couple of hours it start drifting for 1 target. I used multiple target for same material with multiple MFC.