is phase transitions synonymous with critical phenomena
I don't know and I don't think there's a consensus.
A book on condensed matter, for example, says:
The term critical phenomena refers to the peculiar behaviour of a
substance when it is at or near the point of a continuous-phase
transition, or the critical point
It would be easier to disprove your statement rather than proving it, just by finding a counterexample. I cannot think of one, though.
How can we define whether two phases are different from each other?
One may distinguish phases based on:
- physical properties, such as compressibility and rigidity (liquids vs gas);
- distinct symmetries associated with each phase, and a symmetry breaking to go from one to the other;
- an unequivocally defined mathematical quantity (an invariant) that has a different values in either phase, such as a topological invariant like a Chern number or a Pfaffian;
- the presence of a phase transition resulting from a gradual change in some parameter (order parameter)
None of these points is satisfactory though. Because one can still change phases via crossovers and not transitions, in which there are no abrupt changes. So once again I think there is no consensus (yet), and it much easier to say what is not a phase transition or a distinct phase, rather than the opposite.
The order parameter is usually temperature (Ising model, melting points, Bose-Einstein condensation), but does not have to be (e.g. quantum phase transitions). Highly recommend this answer for quantum phase transitions.
The special case of Anderson localisation and other disorder-induced phenomena
When comparing two phases, one usually assumes that both of them are at equilibrium. That is, they had time to settle, and re-arrange their microscoping structure. I could just melt everything and then quench it, freezing its messy internal configuration, and calling it a new "phase" because it looks different than before (this is actually what a glass is).
And usually by equilibrium one talks about thermal equilibrium: that is, let's go over the transition slowly enough that the whole system can thermalise, and there are not small pockets or the other phase that survive (see Kibble-Zurek mechanism).
In this context, disorder-induced "phases" such as an Anderson insulator are tricky. Because, by definition, a disorder-induced phase does not thermalise, thus it cannot reach thermal equilibrium. So you cannot use temperature as the order parameter, and cannot look for a discontinuity in thermodynamic quantities and/or in the free energy.
Hopefully, though, we can agree that the phases "look different" in the Anderson localised and extended regime. Hence why I, personally, still refer to them as distinct phases. Turns out that non-random disorder (Anderson is random) does give you a critical disorder strength for the transition, reminiscent of "usual" transitions with an order parameter (e.g. quasiperiodic disorder).