The basis for boiling point is the saturation pressure of the refrigerant. Saturation is achieved in a closed container where the vapor and the liquid phases are in equilibrium. Thus the pressure of a cylinder of refrigerant, or propane, or any other compressed volatile, will vary with the temperature according to the sat chart....

Add another gas, you get partial pressures in the vapor phase. Thus mixtures of refrigerants yield vapor pressures different from either of the parent components.

And if you make a puddle of ammonia and let it basically sit still, the open space immediately above the puddle will head toward being 100% ammonia vapor. Before it gets there, the temperature of the liquid in the puddle will dip below the boiling point of the refrig because the vapor pressure above the liquid is a partial pressure of ammonia, the rest of it being air pressure. Obviously, that partial pressure is lower than atmospheric. How far it dips depends on the agitation of the gas mix immediately above the puddle, the temperature of the "second" gas and the heat capacitance and heat transfer from the "bottom" side of the puddle. Which means that if your puddle resides on top of 200 mm of concrete floor and the air is basically still above the puddle, the whole assembly drifts slowly down to minus 33 C and basically parks there....

If you want to: try it with a little refrigerant in a pyrex pie plate sitting on top of a slab of insulation. Keep it out of the breeze, it will probably go to minus 37 or 38 C. Fan it with fresh air for a few seconds and the temperature will dip, each time you agitate it.

The jet of flashing liquid thing is accurate though difficult to reproduce. The low temperature front will be near the outer edge of the "definitely wet" portion of the spray. I have done this with butane; some practical limitations to trying to do it with ammonia.

And yes: You can do some very quick damage to portions of a circuit by running "first fill" as liquid to an otherwise warm and deeply vacuumed space. Our procedures call for the vacuum to be broken with vapor from the corresponding portion of circuit or tanker or cylinder. One of the procedures that always gets me is the folks who furnish condensing units with the charge in the HP receiver....Lots of liquid solenoid valves get damaged when the liquid line gets opened up to 100-micron vacuum.