Very little beyond three things.

All occur naturally.

All have negligible global warming potential.

None faces a phase-down timetable.

Their constraints are completely different.

Treating them as one category is the usual error.

Pressure, above everything else.

Systems run at several times conventional pressure.

Above a modest temperature the cycle goes transcritical.

That is a different cycle with its own controls.

Components and pipework must suit the pressures.

Warm climates affect its efficiency more.

Toxicity, not performance.

Thermodynamically it is outstanding.

It requires machinery rooms and detection.

Ventilation, trained operators and safety cases follow.

That suits large industrial sites.

It is rare in occupied buildings for the same reason.

They are straightforwardly flammable.

Not mildly flammable like A2L options.

Charge limits per circuit are strict.

That pushes them toward self-contained appliances.

Small systems suit them very well.

Large distributed systems generally do not.

Yes, and copper is the key one.

Ammonia attacks copper and its alloys.

Systems are built in steel instead.

That affects pipework and components throughout.

Semi-welded plate exchangers suit it well.

Design around it from the start.

No phase-down timetable at all.

Global warming potential is negligible.

No quota-driven price exposure.

Equipment installed now stays serviceable.

That matters over a twenty year life.

It is the strongest case for choosing them.

The building and the maintenance regime.

Machinery room provision where required.

Detection and ventilation appropriate to the substance.

Technician training and certification.

Emergency procedures suited to the hazard.

None of this can be retrofitted cheaply.