Metastability and persistence

A state can persist without being the equilibrium winner

A sample made under high pressure can retain its original structure after the pressure is released. A structural measurement still recognises it a day later. That observation asks two different questions: why is this structure still present, and is it the lowest-free-energy state under the new conditions?

A metastable state occupies a local free-energy minimum but another accessible state has lower free energy. Leaving the local minimum requires crossing a barrier. Persistence depends on the barrier, temperature, defects and available transformation pathways.

Use the same system and constraints when comparing states. Here the composition is closed to exchange, temperature and pressure are fixed, and Gibbs free energy G describes the allowed alternatives. “Energy and the meaning of stability” explains this choice. The comparison must include the relevant competing structures and phase combinations; a low value among a few guesses does not establish that every alternative has been considered.

A local minimum means that nearby allowed changes raise the free energy. A global minimum is the lowest value among all states admitted by the model and constraints. A state can therefore resist small changes yet have a lower alternative farther away. “Equilibrium tendencies and transformation rates” separates that equilibrium comparison from the rate of reaching the alternative.

Read a well, a crest and a lower well

The diagram below makes the distinction concrete. Its configuration coordinate q labels progress through a sequence of atomic arrangements. It is dimensionless, not distance across a specimen and not elapsed time. The vertical values are relative Gibbs energies in arbitrary units for one invented model at common conditions; they are not measured energies of a named material.

A lies at q = 1 with G = 4. Small changes in q to either side raise G, so A is a local minimum. B, at q = 5 with G = 0, is another local minimum and the lowest state in this entire drawn model. A is metastable relative to B; B is the equilibrium-stable state within the model. In a real material, a single drawn path would not establish that no lower state exists elsewhere.

C, at q = 3 with G = 10, is the crest between them. A small change in q to either side lowers G, so C is unstable along this coordinate. Being stationary at an exactly balanced crest is not the same as resisting a disturbance. A single coordinate can reveal an unstable direction, but a minimum on that coordinate alone does not test stability against every other possible disturbance.

For the shown route A to B, the barrier is the rise from the starting well to the crest: 10 − 4 = 6 units. The overall change is the final value minus the initial value: 0 − 4 = −4 units. These are different differences. Starting at B, the reverse barrier along the same route is 10 − 0 = 10 units. Shifting every plotted energy by a common constant would change none of these differences.

Why a lower state need not appear immediately

Creating a region of a new phase can require an initial cost even when a large amount of that phase would lower the free energy. For example, a small embryo introduces an interface with the surrounding phase. Its interfacial cost can outweigh its initial bulk free-energy benefit. Rearrangement and movement of atoms can also be slow. This provides a physical reason for a barrier rather than treating the diagram as a hill that the whole specimen literally rolls over.

A suitable surface or existing interface can provide a different nucleation route. Its barrier need not be the barrier in a uniform interior. Thus knowing only the free-energy difference between A and B does not identify the transformation route or its speed. A diagram for one chosen route also does not establish that it is the route with the smallest barrier.

Local mechanical stability asks whether small allowed displacements or deformations have a restoring tendency about a balanced structure under specified loading constraints. In a simple conservative mechanical model, a potential-energy well confines small disturbances; a crest does not. With no damping the response can oscillate rather than settle immediately. A thermodynamically metastable phase can still be locally mechanically stable.

Keep the tests distinct. A calculation showing an energy increase for every allowed small mechanical deformation establishes local mechanical stability within that model. It does not compare all other phases. A brief small-load test examines the modes and conditions actually probed. Neither result alone establishes global equilibrium or a storage lifetime; mechanical stability is also not a measure of how much load a part can safely carry.

Persistence is conditional

A metastable material can remain useful for a long time under one environment and transform quickly under another. Distinguish thermodynamic stability, mechanical stability against a small disturbance, and a measured lifetime. Demonstrating a state briefly does not establish that it survives storage, decompression or repeated use.

Consider the invented retention records in the table. Separate matched samples are made in the same initial phase, pressure is released, and each is then held at its stated temperature in the same inert surrounding gas with no applied mechanical load. The structural method samples specified regions at the listed times. For this example it can identify a competing phase when its sampled volume fraction reaches 1%; the stated observations have adequate resolution to distinguish the reported 15% result. These are supplied teaching observations, not data for a named phase.

At 300 K the initial phase is identified immediately after release, at 1 h and at 24 h, with no competing phase detected above the stated limit. Report what was seen at those times and regions. This does not exclude a smaller transformed fraction, an unsampled region or a later change. At 350 K a matched sample already contains 15% competing phase at the 10 min observation. Its exact transformation onset between observations is not given.

These records show that retention cannot be transferred unchanged between the two tested conditions. They do not determine the free-energy ranking at either temperature. Temperature can change both the relative free energies and the accessible rates. Faster atomic movement can compete with a changed transformation driving force, so “hotter always transforms faster” is not a general law.

A useful property needs its own endpoint and evidence during the intended interval. Recognising a phase at 24 h does not by itself show that its conductivity, strength or another desired response stayed within a requirement. Keep forming a state, observing its retention and establishing its useful performance as separate claims. “Temperature, moisture and chemical surroundings” develops further environmental changes.

Worked examples

Try the reasoning

1. Does metastable mean imaginary?

   Show the answer and reasoning

   No.

   A real observed state can be metastable.

2. Can changing temperature shorten persistence?

   Show the answer and reasoning

   Yes.

   It can make a transformation pathway more accessible.

3. A different invented model has a local minimum U at 7 energy units, a lower minimum V at 2 and a crest at 11 on the route between them. A second allowed route has its crest at 9, with the same endpoints. Calculate the forward barriers and net change. Does the second route make V thermodynamically more favourable?

   Show the answer and reasoning

   The forward barriers are 11 − 7 = 4 units and 9 − 7 = 2 units. The net U-to-V change is 2 − 7 = −5 units for both. The alternative route changes the barrier, not the endpoint ranking.

   Subtract the starting well from each crest, not zero. U is metastable relative to V under the stipulated common constraints. A smaller barrier can alter accessibility, but these invented numbers alone still give no transformation time or guaranteed dominant route.

4. A candidate crystal is calculated to resist every allowed small mechanical distortion in a stated model, yet another structure has lower Gibbs free energy at the same temperature, pressure and composition. A sample also retains the candidate phase at two observations one hour apart. Which kind of stability has each result tested?

   Show the answer and reasoning

   The first result establishes local mechanical stability within the model. The second comparison shows the candidate is not the global equilibrium winner; if it occupies a local free-energy minimum separated by a barrier, it is metastable. The observations establish retention at the measured times, not an indefinite lifetime.

   Local restoring response, a comparison of alternative phases and finite structural observations answer different questions. The stated evidence does not by itself quantify the barrier or prove stable response to every unmodelled disturbance in a real specimen.

5. A high-pressure phase is identified in one sampled region immediately after release and again after 24 h in dry gas at 300 K. Its desired electrical response was measured only before release. A proposed use requires 30 days in humid air. What can be retained from the evidence, and what missing comparison matters?

   Show the answer and reasoning

   Retain the finite structural observations under dry-gas, 300 K conditions. They do not establish 30-day humid-air retention or the post-release electrical response.

   Formation pressure, retention environment and a functional endpoint are separate parts of the claim. Examine representative structural and electrical responses after release over the relevant exposure, with the method’s detection and uncertainty limits. A preserved phase label alone cannot substitute for property evidence.

What to carry forward

• Local persistence and global equilibrium are different.
• Attach conditions and duration to retention claims.
• Read barriers from the starting well to the relevant crest; the net endpoint change is a different quantity.
• A local minimum on one path does not test every mechanical mode or every competing phase.
• A finite structural record needs its conditions, sampled region, detection limit and observation times; it does not by itself establish a useful-property lifetime.