Abstract
Yu et al. suggested calculating precisely the size ranges of the three parts of our figure 3A, adjusting the free-energy levels in figure 3B, and considering the shape effect in the first-principles calculation. The first and second suggestions raise strong concerns for misinterpretation and overinterpretation of our experiments. The original calculation is sufficient to support our claim about crystalline-to-disordered transformations.
Introduction
The birth of a crystal—nucleation—is a fundamental process in chemistry and physics, yet observing it at the atomic scale remains a significant challenge. In 2021, Jeon et al. published a study in Science demonstrating that tiny nanoclusters do not simply grow; they fluctuate between crystalline (ordered) and disordered states. A follow-up comment by Yu et al. challenged these findings, arguing for discrete "magic number" size ranges, a definitive thermodynamic stability of the disordered state at small sizes, and the inclusion of specific geometric "shape effects." This response defends the original "gradual" model of nucleation against these critiques, asserting that the transition is a continuous process influenced by the surrounding environment.
Statistical Rebuttal: The Danger of Smoothing
A primary point of contention involves the statistical treatment of experimental data. Yu et al. proposed identifying specific size ranges for transitions. However, Jeon et al. argue that such an approach risks "overinterpreting" the data. By applying statistical smoothing to create better curve fits, one may achieve higher correlation coefficients ($R^2$), but at the cost of obscuring real physical deviations and the stochastic nature of atomic movement. The authors invoke Simpson’s Paradox to illustrate how "averaging averages" can lead to misleading conclusions, emphasizing that the observed "scatter" in data represents the actual random behavior of atoms at the nanoscale.
The Role of Adatoms and Environment
A key insight of the original work is that nanoclusters are not isolated systems. They exist in an environment of mobile adatoms (loose atoms on the surface). The interaction between these adatoms and the cluster provides the energy necessary to trigger transitions between ordered and disordered states.
The authors maintain that their empirical fitting was intended as a "guide to the eye" for a continuous process influenced by these environmental factors, rather than a map of discrete physical boundaries.
Thermodynamic Landscapes and Size Increase
The thermodynamics of these transitions change gradually as the cluster grows. The "dataset" in question consists of high-resolution Transmission Electron Microscopy (TEM) movies and the resulting Temporal Fractions of the Crystalline State (TFCS)—a measure of how much time a cluster spends in an ordered state.
While critics suggested that the disordered state is definitively more stable at small sizes, Jeon et al. argue that experimental limitations, specifically the 10ms temporal resolution of TEM, make it difficult to verify absolute stability values. Any crystalline state existing for less than 10ms would be missed, leading to an artificial underestimation of the crystalline fraction.
Furthermore, the authors demonstrate that accounting for polyhedral shapes (the "shape effect") would only strengthen their original claim: adatom binding provides sufficient energy to drive the observed disorder-order transitions.
Conclusion and Future Outlook
The authors conclude that their original, simpler models provide the most honest representation of the experimental data without falling into the trap of overinterpretation. While the absolute stability of the disordered state in very small clusters remains a "controversial problem," it presents an opportunity for future in-depth studies. Such work may require even faster imaging techniques or more isolated experimental environments to resolve the stochastic nature of nucleation fully.
Summary updated