The air in the room is about 78% Nitrogen, 21% Oxygen and 1% Argon. Once you replace 95% of that with CO2 you will have:
95% CO2
3.75% Nitrogen
1.05% Oxygen
0.05% Argon
(percentage in number of molecules)
That's high on Oxygen and Nitrogen and low on Argon. You could burn some carbon in the box to convert most of the Oxygen to CO2 and add some argon before bringing the percentage of CO2 up to 95%.
If you want to remove every bit of atmosphere before building a new atmosphere from scratch it will be MUCH more expensive.
Neat. What temperature do you get when lowering the pressure from 1000 millibar (1 Bar) at 20°C (293.15°K) to 6 millibar at constant volume? Is it close to the 173°K (-100°C) that the OP desires?
Treating this as a pure adiabatic expansion, starting with 1 liter of air at 1000 mBar/300K, I get a final volume of 38 liters at 6mBar/70K (about -200 C). (Edit: corrected)
Of course this will be slow with a pump, and not adiabatic, as the chamber housing will transfer heat from the surroundings to the gas in the chamber. Furthermore if the air is humid, water will condense and keep the temperature higher than calculated.
Good exam question for an undergraduate thermal physics or chemistry class!