When a crystalline film grows on a substrate, the atoms face a choice. They can orient themselves to minimize the film's own surface energy — free epitaxy, where the crystal chooses its preferred direction regardless of what lies beneath. Or they can lock onto the substrate's crystal lattice, sacrificing internal preference for interfacial bonding — locked epitaxy, where the substrate dictates orientation. Most material systems fall clearly into one regime or the other. The choice is made by thermodynamics before the first atom lands.
Fe₄N grown on mica sits on the boundary. The surface energy cost of locking onto mica's lattice is almost exactly balanced by the interfacial energy gained from chemical bonding. The system is thermodynamically frustrated — neither regime wins decisively. In free epitaxy, Fe₄N grows with its (001) face up. In locked epitaxy, it grows (111). Two distinct crystal orientations, selected by which energy term dominates by a margin too small for the material's own chemistry to resolve.
The researchers found that shining light on the interface during growth resolves the frustration. Photo-excited carriers act as chemical potentiators, enhancing the interfacial affinity between film and substrate. The locking criterion — the ratio of interfacial energy gain to surface energy cost — crosses its critical threshold. The system switches deterministically from free to locked epitaxy. Turn the light off, and the next film grows free. The light is not forcing a new state into existence. It is selecting which of two pre-existing states wins a competition that was too close to call.
The key is that the system must be frustrated for the switch to work. A strongly coupled interface is already locked; light cannot unlock it. A weakly coupled interface is already free; light cannot push it far enough to lock. Only at the boundary — where the energy balance is marginal — does a small perturbation choose the outcome. The frustration is not a problem to solve but a condition that enables external control. A decisive system cannot be switched. An indecisive one can be switched by almost anything.
The general principle: controllability requires proximity to a threshold. A system deep in one regime is stable but unresponsive. A system at the boundary between regimes is sensitive to perturbations that would be negligible elsewhere. The light does not provide the energy for a new crystal orientation — it tips a balance that was already even. Design for frustration, and you design for control.