Compeletely purging a milling cell from atmosphere is problematic indeed. Solution depends on number of prenumatic connections with argon.

I used a ring mill with 'single' pneumatic connection and I only diminished the oxidation. But I couldn't compeletely eliminated. I used 5 bar of argon. I fill the chamber, closed the valve, removed the pneumatic hose, open the valve to release the gas inside of chamber, close the valve immediately, put pneumatic hose again, apply 5 bar of argon and fill the chamber. I repeated this procedure 5 times for my purpose. It decreased the oxidation to negligible levels but it didn't compeletely elimate it. If you don't require to work with a lot of samples like mysely, I suggest repeat the cycle 10 or 15 times. If your single connection can reach every part of chamber, I mean such as over filling of your chamber with too much sample prevent argon gas to reach everywhere, this procedure will decrease oxygen amount 20 % of it each time. Therefore, 5 cycles mean decreasing atmosphere to 0.032 % of volume.

If your grinding cell has two connections, one inflow and one outflow, I think problems might occur due to excessive sample mass and/or small deadend spots decrease flow velocity. Perhaps purging longer durations with this system and opening and closing the flow to allow diffusion to influence those deadend locations might solve your problem.

I hope this answer help you with your problem.