A
phase change is more sensitive to temperature. The second choice would be a gas. A solid or a liquid are much worse. That's why usual engines uses gases or vapours, not solids.
If you can have a hermetic enclosure for the vapour,
evaporation makes a huge volume change. Mixtures evaporate over a temperature range.
Melting can make a usable volume change.
Paraffins are known for that and used at thermostats for room heaters. Melting occurs over some degrees range. The paraffin's mixture composition controls independently where melting begins and ends.
https://en.wikipedia.org/wiki/Wax_thermostatic_elementMaybe you can even buy such a thermostat and integrate it in your creation?
If you must stick at a liquid or a solid without melting, substances expand more when they're near to their melting or boiling point. Paraffins are again a choice. Or consider something like polyethylene, or stretched rubber (but it's less durable). Some less common solids have transitions that are not melting, and the volume can change a bit more there.
Some odd materials, especially
shape memory alloys, do have dimensions abnormally sensitive to temperature. Expensive, not very convenient, but they're usable.
https://en.wikipedia.org/wiki/Shape-memory_alloyThis is more exotic and still fashionable than melting paraffin, it has some "wow" contents.
In good cases, expansion is few per-cents, so usually the
shape of your construction must amplify it. For instance mercury thermometers have a flask where a significant volume expands, and a narrow tube where little volume makes a big change in height. The same for paraffin thermostats, where some paraffin volume pushes on a narrower needle to regulate the throughput of heating water. If using metals (say, to resist flame temperature) one makes a bimetallic strip of them so the movement gets usable.
https://en.wikipedia.org/wiki/Bimetallic_stripThe same could be done with polymers. Interesting choice if it fits the art, like petals opening at warmth.
Ferromagnetic materials (and ferrimagnetic for hair-splitters) become non-magnetic above their "
Curie temperature"
https://en.wikipedia.org/wiki/Curie_temperatureSome materials were developed with a transition around room temperature, over a not-too-wide range. A permanent magnet would then attract these materials only when the room is colder. Not my first choice neither, because a design is hard to compute and predict. About as difficult to use as shape memory alloys.