Disorder usually destroys superconductivity. Random impurities scatter electron pairs, breaking the quantum coherence that allows current to flow without resistance. Anderson's theorem protects conventional superconductors against non-magnetic disorder, but once disorder exceeds a threshold, superconductivity collapses. The relationship between disorder and superconductivity is understood to be adversarial.
Remeika-type quasiskutterudites (R₃M₄Sn₁₃ and R₅M₆Sn₁₈, where R is a rare earth and M is a transition metal) don't follow this story. Increasing atomic-scale disorder — introducing dopants, disrupting the lattice — creates locally superconducting regions with critical temperatures higher than the bulk material's transition temperature. The bulk transition temperature and the local critical temperature diverge as disorder increases, and the divergence is largest at entropy maxima. Maximum disorder produces maximum local enhancement.
The mechanism involves a competition between two effects that operate at different scales. Locally, disorder modifies the electronic environment in ways that strengthen pairing — the same scattering that disrupts long-range coherence can enhance short-range coupling. Globally, disorder prevents these locally enhanced regions from establishing coherent phase relationships across the material. The result is percolative superconductivity: superconducting islands that may or may not form a connected path through the material, depending on temperature and dopant concentration.
Upper critical field measurements confirm this picture directly. Two distinct branches of H_c2(T) appear — one corresponding to the bulk transition and one to the local superconducting regions. The material isn't uniformly superconducting with higher noise; it contains two distinct superconducting populations with different properties.
The thermodynamic signature is precise: entropy maxima coincide with the largest separation between bulk and local critical temperatures. This is not an accidental correlation — it means the parameter that maximizes microscopic randomness also maximizes the difference between local and global superconducting behavior. The mess is not incidental to the enhancement. The mess is the enhancement.
This inverts a default assumption about optimization. In most systems, the path to better performance runs through reducing disorder — purer materials, cleaner interfaces, more controlled fabrication. Here, the system performs better locally precisely because it performs worse globally. The global coherence suppression is not a side effect of disorder that happens alongside the local enhancement. The same disorder that breaks global coherence creates the local electronic environments where pairing is stronger. You cannot have the enhancement without the mess.
The finding has a structural analogue in any system where local optimization and global coordination pull in opposite directions. A market where individual traders exploit local mispricings but can't coordinate a global strategy. An ecosystem where disturbance creates niches that support higher local biodiversity at the expense of system-wide stability. A research program where disorder in methodology produces occasional breakthroughs that a uniform approach would miss. The principle is the same: when the mechanism that enhances local performance is the same mechanism that disrupts global coherence, you face a genuine tradeoff, not a problem to be solved.