Probing Anharmonic and Heterogeneous Carrier Dynamics Across Sublattice Melting in a Minimal Model Superionic Conductor Sucharita Niyogi, Takenobu Nakamura, Genki Kobayashi, Yasunobu Ando, Takeshi Kawasaki Abstract Despite decades of research, the microscopic origin of sublattice melting and fast ion transport in superionic conductors remains elusive. Here, we introduce a chemically neutral minimal binary model consisting of a rigid host lattice stabilized by short-range steric repulsion and a soft carrier sublattice interacting via long-range Wigner-type forces. This contrast naturally produces distinct melting temperatures and an intermediate sublattice-melting phase in which carriers become fluidlike while the host remains crystalline. Molecular-dynamics simulations identify three dynamical regimes—crystalline, sublattice-melt, and fully molten—marked by sharp changes in diffusivity, structural correlations, and dynamic heterogeneity. Near sublattice melting, carrier motion is strongly anharmonic and spatially heterogeneous, beyond mean-field hopping descriptions. By tuning the density, we demonstrate that sublattice melting can be continuously controlled, establishing a direct link between lattice softness, anharmonicity, and collective ion transport. This work provides a unified microscopic foundation for designing mechanically robust, high-performance superionic conductors operable near ambient conditions. Keywords: Solid-state batteries, carrier transportation, sublattice melting, dynamic heterogeneity, anharmonicity. I Introduction The rapid advancement of energy storage technologies has placed rechargeable batteries at the center of modern society, powering applications from portable electronics to electric vehicles and large-scale renewable energy integration. At the core of all such battery technologies is ionic transport: ionic conductors, whether liquid or solid, are essential for efficient charge transport and directly determine battery performance, longevity, and safety. While liquid electrolytes in conventional Li-ion batteries provide high ionic mobility and performance 10 , these volatile solvents are flammable, chemically reactive, and prone to leakage, making them a primary source of thermal runaway and catastrophic failure in high-energy-density devices 38 . Therefore, overcoming these limitations is crucial for the safe and reliable deployment of next-generation batteries. In this context, solid-state electrolytes, particularly superionic conductors 7 ; 21 ; 40 ; 41 ; 46 ; 56 ; 60 ; 4 ; 44 ; 28 ; 9 ; 26 ; 20 ; 51 ; 23 ; 16 ; 61 , constitute a distinct class of materials in which ions move almost as freely as in molten salts while the host lattice retains long-range crystalline order. Their defining feature is the coexistence of two distinct sublattices: an immobile framework that maintains structural integrity and a mobile-ion sublattice that becomes disordered at elevated temperatures 45 . The temperature dependence of this order-disorder transition varies widely among materials. For example, NaCl exhibits a pronounced increase in ionic conductivity only near its melting point, consistent with conventional melting behavior, whereas some superionic conductors, such as PbF 2 , show a gradual enhancement of conductivity over several hundred kelvin well below the melting temperature 29 ; 23 ; 16 . Others, including AgI, undergo a first-order phase transition accompanied by an abrupt increase in ionic conductivity by more than three orders of magnitude. Similar melting-like transport behavior has also been reported in experiments on other charge carriers, including hydride ions (H - ), indicating the generality of this phenomenon across ion species 51 . Despite these observations, the microscopic origin of such an anomalous transition remains unresolved. Some recent analyses of layered and tunnel-type superionic conductors have shown that carrier transport is strongly influenced by many-body interactions among mobile ions, reshaping the diffusion landscape beyond simple single-particle hopping; here, Raman measurements and MD simulations further show that short-range cation-cation repulsion modifies activation barriers, underscoring the collective nature of fast-ion motion 27 . Together, these studies suggest that sublattice melting and strong anharmonic lattice fluctuations are both essential ingredients for superionic transport; however, a direct microscopic link connecting these phenomena remains to be clearly established. To place these observations in context, sublattice melting 18 ; 40 ; 57 ; 5 ; 4 ; 25 refers to a state in which the mobile ionic network becomes dynamically disordered while the host framework remains crystalline, enabling liquid-like conductivity without structural collapse. Early mean-field models 57 attributed this transition to a balance between defect formation energy and configurational entropy, predicting disorder within one ionic sublattice prior to complete melting. However, their defect-based mean-field framework precludes access to real-space ion dynamics, spatial correlations, and anharmonic lattice effects central to superionic transport. Experimental studies on CuI provided early dynamical evidence for sublattice melting: NMR measurements showed anomalously enhanced relaxation near the superionic transition, consistent with a molten copper sublattice within an ordered iodide framework 5 ; 4 . Subsequent theoretical treatments 18 described the precursor defect proliferation using a cube-root dependence of the defect chemical potential, reproducing the premelting characteristics observed in AgI and PbF 2 . More recently, analogous phenomena have been realized in asymmetric colloidal crystals 40 , where smaller charged particles delocalize within a crystalline matrix of larger oppositely charged spheres, and in ultrasmall copper selenide clusters exhibiting liquid-like cationic sublattices even at ambient temperature 58 . Complementary experimental studies have since directly visualized this selective sublattice melting 11 , reaffirming its universality in ion-conducting solids. Despite these advances, the microscopic origin of such selective disorder remains poorly understood, as most existing descriptions rely on mean-field or defect-based frameworks that neglect correlated particle motion and its feedback on the lattice vibrational landscape. Selective sublattice disorder reshapes the vibrational potential energy landscape, introducing anharmonicity that enables nonlinear energy exchange and mode coupling between phonons 22 ; 59 ; 6 . Such anharmonicity effectively softens the lattice and lowers local activation barriers for ionic motion. Early studies on superionic AgI 22 proposed that Ag + ions undergo strongly anharmonic, semi-liquid-like thermal vibrations that extend toward neighboring interstitial sites. This anharmonic motion can overcome local potential barriers and promotes ionic hopping, while appearing in diffraction as a pseudo-static occupation of interstitial sites rather than true site disorder. High-pressure analyses 59 further showed that harmonic or mean-field descriptions fail to capture melting behavior driven by such vibrations. More recent first-principles and spectroscopic studies on sodium-ion conductors 6 demonstrated that strong host-ion anharmonic coupling induces order-disorder transitions with soft modes persisting across the phase change, directly linking lattice dynamics to fast ionic conduction. Despite these advances, most theoretical models neglect correlated dynamics within the mobile sublattice, leaving the interdependence of these effects largely unexplored. Together, these studies suggest that sublattice melting and anharmonicity are not independent phenomena but mutually reinforcing aspects of the same underlying physics, in which selective ionic disorder feeds back into lattice softening, and vice versa. In particular, recent work on Li 10 GeP 2 S 12 has shown that Li + conduction proceeds via correlated and cooperative migration of densely packed ions 60 , explicitly invoking the concept of dynamic heterogeneity through the coexistence of fluid-like, highly mobile regions and immobile rigid domains. Such spatially inhomogeneous dynamics reflect a strongly anharmonic environment, where vibrational instability and sublattice disorder reinforce one another. This coexistence of mobile and immobile regions is a well-known hallmark of glass-forming systems and other dynamically heterogeneous materials. More broadly, dynamic heterogeneity is a central organizing principle in the physics of glass-forming systems, where it has been extensively studied as the microscopic origin of transport anomalies, violation of the Stokes-Einstein relation, and cooperative motion 37 ; 62 ; 63 ; 2 ; 8 ; 30 ; 3 ; 31 . At the microscopic level, such heterogeneity has been directly linked to anharmonic vibrational fluctuations that reveal highly localized, anomalously soft structures associated with quasilocalized, nonphononic vibrational modes 65 . These soft regions show fat-tailed stiffness distributions and underlie spatially heterogeneous dynamics; in strained amorphous solids, low-frequency modes soften and localize near instability, predicting collective rearrangements 53 . In this light, dynamic heterogeneity offers a natural conceptual bridge between sublattice melting and anharmonicity in superionic conductors: spatial variations in local anharmonicity amplify mobility contrasts, giving rise to cooperative, string-like dynamics beyond mean-field descriptions, as observed in glass-forming liquids where enhanced vibrational fluctuations precede collective string-like rearrangements 32 . The extent to which analogous vibrational-heterogeneity couplings govern fast-ion transport in crystalline solids remains an open question. To elucidate this connection, we employ a minimal and tractable model to systematically probe how temperature and interactions control sublattice melting and collective carrier motion. This approach allows us to identify the conditions under which liquid-like regions emerge within an otherwise rigid lattice, quantify correlated ion dynamics beyond mean-field descriptions, and explore the possibility of achieving superionic conduction near room temperature . By linking lattice softness, anharmonicity, and dynamic heterogeneity, this framework provides a tractable route to understand fast-ion transport and design mechanically robust, high-performance superionic conductors. II Results Our analysis focuses on three key mechanisms—selective sublattice melting, structural heterogeneity, and anharmonicity—to elucidate ion-transport dynamics in two-dimensional systems. In two dimensions, the true long-range translational order is prohibited at finite temperature by the Hohenberg-Mermin-Wagner theorem 43 ; 47 ; 49 ; 48 , although the orientational order may persist. Our conclusions do not rely on strict translational long-range order; rather, they concern the emergence of liquid-like transport within an ordered host through enhanced anharmonicity and spatially heterogeneous, string-like motion. Two-dimensional systems are therefore used primarily for clarity and visualization, while corresponding three-dimensional results are presented in the Supporting Information. II.1 Microscopic Configurations and Sublattice Disorder To investigate the microscopic dynamics of the two-dimensional host-carrier system, we employ model systems in which the host and carrier form stable crystals at low temperatures. The choice of such a system reflects the long-range nature of the interaction potential, analogous to the Coulomb force (see “Materials and Methods” for details). Under these conditions, low-density carriers naturally organize into ordered patterns at low temperatures, forming a well-defined crystalline state. As the temperature increases, the host lattice remains crystalline while the carrier sublattice progressively melts. This configuration provides a mechanically stable reference state from which the subsequent thermal evolution and melting processes can be systematically explored, ensuring that the high-temperature dynamics originate from physically meaningful particle arrangements established at low temperatures. Figure 1: Structural representations of the host-carrier system: ( A ) Schematic representation of the interaction scheme between interparticle pairs: the host-host interaction length is denoted as \sigma_{\mathrm{H}}=\sigma_{\mathrm{HH}} , the host-carrier interaction length as \sigma_{\mathrm{CH}} , the non-additive interaction length \sigma_{\mathrm{CC}} between carrier-carrier is represented by the red dotted line, and the effective-carrier interaction length is chosen to be \sigma_{\mathrm{C}} (see Methods for more details). ( B ) Low-temperature configuration obtained using a Wigner-type interaction, illustrating the ordered host lattice (blue) with carrier particles (red) occupying energetically favorable interstitial sites. ( C–F ) Representative snapshot configurations illustrating crystalline, sublattice-melt, and fully molten states across increasing temperature (top to bottom). For clarity, each snapshot includes an inset highlighting a magnified region of the configuration, emphasizing local structural arrangements and carrier environments. Figure 2: Selective sublattice melting and structural evolution: ( A ) Schematic representation of the diffusivity D_{\alpha} as a function of inverse temperature 1/T , illustrating three distinct regimes: (I) crystalline, (II) sublattice melting, and (III) full melting. The characteristic temperatures T_{m}(\sim 0.35) , T_{f}(\sim 0.50) , and T_{l}(\sim 3.50) are indicated, corresponding respectively to the onset of sublattice melting, the freezing of carrier motion, and the liquid-like regime. ( B ) Representative snapshots illustrating structural evolution across these regimes. ( C ) Radial distribution functions g_{\alpha\alpha}(r) ( \alpha\in\{{\rm C,H}\} , where C : Carrier and H : Host) corresponding to each regime. Region II shows disordered carrier sublattice coexisting with ordered host lattice, evidencing selective carrier melting. Figure 3: Trajectory of carriers at different time intervals and temperature regimes: Panels ( A–F ) show carrier trajectories within region II-III of the diffusion plot (as shown in Fig. 2 ), corresponding to the sublattice melting regime. Panels ( A, B ) present trajectories at the melting temperature of host within region III, beyond the sublattice melting regime, while panels ( C, D ) correspond to the higher-temperature side of the sublattice melting regime (region II). Panels ( E, F ) illustrate more heterogeneous behavior in the trajectories on the lower-temperature side of region II. Results are shown for packing fraction \varphi=0.85 . Particle identities are color-coded as indicated by the accompanying colorbar. Figure 4: Microscopic dynamics and quantify the extent of dynamical heterogeneity: ( A-C ) Carrier dynamics are characterized by the mean-squared displacement (MSD), the self-intermediate scattering function F_{s,\mathrm{C}}(q,t) , and the four-point susceptibility \chi_{4,\mathrm{C}}(t) for the system of packing fraction \varphi=0.85 at different temperatures (please see the top color bar). The characteristic temperatures T_{m}(\sim 0.35) , T_{f}(\sim 0.50) , and T_{l}(\sim 3.50) —marking the onset of sublattice melting, the freezing of carrier motion, and the liquid-like regime, respectively—are indicated by red circles, blue squares, and black triangles connected with dotted lines. ( D-E ) Inverse-temperature dependence of the Stokes-Einstein (SE) ratios D_{\rm{C}}\tau_{\alpha,\mathrm{C}} and D_{\rm{C}}\tau_{\chi_{4,\mathrm{C}}} , highlighting pronounced SE violation at low temperatures ( E ). A guideline in ( D ) emphasizes the slope change near the sublattice-melting transition. Figure 5: Interplay between dynamic and static heterogeneity near the sublattice-melting point: ( A ) Spatial map of mean square displacement over t\in[t_{0},\,t_{0}+\tau_{\alpha}] at T=0.45 , showing coexisting liquid-like and immobile crystalline domains. Displacement vectors in ( A ) further reveal string-like cooperative motion and hopping pathways within liquid-like regions, contrasted with localized, heterogeneous vibrations in crystallized domains. ( B ) Static configuration at t_{0} , coloured by nearest carrier-carrier distance, revealing anharmonic regions that align with the dynamically mobile zones in ( A ). Together, the two panels show that dynamical heterogeneity emerges from underlying static distortions of the carrier sublattice. Figure 1 illustrates the structural framework underpinning our model. The host and carrier particles interact through non-additive pair potentials that define distinct effective length scales, as summarized in Fig. 1 A . At low temperature (Fig. 1 B ), these interactions stabilize a crystalline arrangement in which carriers occupy well-defined interstitial positions within the host lattice—a reference configuration from which all subsequent temperature-dependent transformations are analyzed. To provide a broader picture, Figs. 1 C–F presents representative structural snapshots at different temperatures, where one can clearly observe the progressive loss of order. In the most stable, low-temperature state ( T=0.30 in Fig. 1 C ), both host and carrier particles exhibit a well-preserved crystalline symmetry. Upon heating, a sublattice melting transition emerges (Figs. 1 D,E ). At the lower bound of this regime ( T=0.50 ), partial disordering of the carrier sublattice is observed—regions of crystalline order coexist with melted domains—while the host lattice remains intact. At a higher temperature ( T=2.00 ), the carrier sublattice becomes fully disordered, yet the host lattice continues to preserve its crystalline arrangement, indicating that both states lie within the sublattice melting regime where only the carrier network loses structural order. At sufficiently high temperatures ( T=10.0 in Fig. 1 F ), both host and carrier particles lose positional order, corresponding to complete melting of the system. To quantify particle mobility and capture the key characteristics underlying this phenomenon, we computed the mean squared displacement (MSD) of the carrier and host particles separately (see Eq. 4 in Materials and Methods and Fig. 4 A ). The long-time diffusive behavior of the MSD was then used to calculate the self-diffusion coefficient D_{\alpha} using the Einstein relation (see Eq. 5 in Materials and Methods for more details). This analysis was performed separately for both carrier and host particles, allowing us to directly compare their respective mobilities and the influence of temperature on their transport dynamics. In Fig. 2 A , we identify four distinct Arrhenius slopes that correspond to different dynamical regimes: (I) a narrow crystalline regime; (I+II) an intermediate, metastable coexistence regime characterized by partial disordering of the carrier sublattice, where melted and crystalline carrier regions coexist and gradually disappear as the system approaches the full sublattice-melting point T_{f}(\sim 0.50) ; (II) broad sublattice-melting region, and (III) a complete melting regime where both particle types are fully disordered (referred by the starting point T_{l}(\sim 3.50) ). In regime (I), the carrier diffusivity exhibits an Arrhenius slope comparable to that of the host, consistent with activated single-particle hopping. By contrast, the two-order-of-magnitude enhancement of carrier diffusivity in regime (II) signals the onset of collective motion, involving correlated groups of carriers rather than independent hops. Despite this large increase in mobility, the carrier diffusivity across region (II) does not show a sharp discontinuity, whereas the host diffusivity displays a pronounced jump near region (III), indicative of a first-order melting transition. These findings are in good agreement with previous experimental observations that reported similar melting signatures in related systems 12 ; 51 . A direct comparison between our model predictions and available experimental trends is provided in the Supporting Information. To further clarify the microscopic origin of the high-temperature transport behavior, we analyze the carrier dynamics well above the host melting point in the Supporting Information for the three-dimensional system. There we show that carrier size has a pronounced impact on transport, especially in the high-temperature regime. While an activated Arrhenius behavior associated with carrier sublattice melting is already present, a key new feature emerges for systems with comparatively larger carriers: at elevated temperatures, the diffusivity exhibits a second Arrhenius regime, which follows the activated form D=D_{0}\,\exp\!\left(-E_{\mathrm{a}}/k_{\mathrm{B}}T\right) , and is directly associated with the onset of host-lattice melting and reflects a distinct, host-controlled activation process (see Supporting Information for details). At still higher temperatures, once the host lattice becomes fully disordered, transport crosses over to a kinetic regime characterized by a rapid, simple liquid-like increase of diffusivity consistent with the Einstein relation D=\left(k_{\mathrm{B}}T/\xi_{\mathrm{eff}}\right) . This sequence of regimes demonstrates how increasing the carrier size qualitatively separates carrier- and host-controlled activation processes. To aid the interpretation of the transport regimes, Fig. 2 B shows magnified cross-sectional views at representative temperatures ( T=0.30 , 2.00 , and 10.00 ), corresponding to the crystalline, sublattice-melted, and fully molten states already discussed in Figs. 1 C, E, F . These views highlight the progressive loss of order in the carrier sublattice while the host lattice remains intact in the sublattice-melting regime, followed by complete disordering of both components at high temperature. To quantitatively characterize these structural correlations, we calculate the radial distribution function (RDF), g_{\alpha\alpha}(r) , for both carrier and host particles (see Eq. 3 in Materials and Methods). As shown in Fig. 2 C , the RDF exhibits sharp periodic peaks at low temperature (region I), reflecting the long-range crystalline order of both the host and the carrier particles. In the intermediate regime (region II), the contrast becomes evident: the RDF of the host particles retains pronounced peaks even at very high temperatures (e.g., T=2.00 ), while that of the carriers broadens into liquid-like correlations, corroborating the sublattice melting identified from the snapshots. At high temperatures (region III), both sublattices lose positional order, and the RDFs of hosts and carriers display the characteristic liquid-like structure with damped oscillations. A detailed analysis of the temperature evolution of the carrier and host RDFs, including the emergence of excess short-range features associated with activated carrier hopping, is provided in the Supporting Information. II.2 Dynamical Heterogeneity in Solid Ionics While our earlier analyses suggest that the system might exhibit significant dynamical heterogeneity, a direct visualization offers deeper insight. To this end, we trace the trajectories of representative carrier particles within the host lattice and map their motion across distinct time windows and temperature regimes. Figure 3 presents the carrier trajectories at representative temperature regimes, illustrating the evolution of dynamic behavior across the sublattice melting transition. At high temperatures like T=7.0 (see Figs. 3 A–B ), both the host lattice and the carrier sublattice are completely melted. Consequently, the trajectories exhibit homogeneous motion at both short ( t=1.50\tau_{\alpha} ) and long ( t=40.00~\tau_{\alpha} ) time windows, where \tau_{\alpha} denotes the structural relaxation time corresponding to each temperature. Upon lowering the temperature (Figs. 3 C-D ), close to but still above the sublattice melting regime, the short-time trajectories remain homogeneous, while at longer times, a distinct honeycomb-like pattern emerges. This pattern reflects the re-establishment of long-lived structural correlations within the host lattice, indicating that the underlying crystalline framework remains dynamically stable even on extended time scales. The most intriguing behavior appears near the onset of the sublattice melting transition (Fig. 3 E ). In this intermediate regime, the system exhibits clear signs of dynamical heterogeneity. As seen more clearly in the inset of Fig. 3 E , certain carriers undergo collective melting and diffuse along interconnected pathways, while others remain temporarily trapped within their local potential wells, unable to hop to neighboring sites. The coexistence of these mobile and immobile regions gives rise to a heterogeneous spatial distribution of dynamics, where the carrier trajectories align with the underlying hexagonal lattice, forming a pronounced honeycomb-like network. This observation confirms that near the sublattice melting transition, the dynamics are governed by spatially correlated heterogeneous motion, characteristic of cooperative rearrangements in partially ordered systems. To elucidate the microscopic dynamics and quantify the extent of dynamical heterogeneity in the system, we analyzed several dynamic observables across different temperature regimes, as summarized in Fig. 4 . The panels A-C respectively show the mean squared displacement (MSD) of the carrier particles, the self-intermediate scattering function F_{s,\rm{C}}(q,t) (see Eq. 6 in the Materials and Methods), and the four-point susceptibility \chi_{4,\rm{C}}(t) (see Eq. 10 in the Materials and Methods). For reference, three characteristic temperatures— T_{m}(\sim 0.35) , T_{f}(\sim 0.50) , and T_{l}(\sim 3.50) —identified from the diffusivity behavior in Fig. 2 , are marked in all panels. These represent, respectively, the onset of sublattice melting, the freezing of carrier motion, and the liquid-like regime. The MSD data [Fig. 4 A ] reveal that at low temperatures near T_{m} , particles exhibit pronounced caged motion, which gradually weakens with increasing temperature and eventually dissolves above T_{f} , leading to fully diffusive dynamics. Correspondingly, the self-intermediate scattering function F_{s,\rm{C}}(q,t) [Fig. 4 B ] exhibits the emergence of a plateau around T_{f} , signaling slow structural relaxation associated with carrier localization. At higher temperatures, F_{s,\rm{C}}(q,t) decays rapidly without any plateau, consistent with liquid-like mobility. The four-point susceptibility for carriers \chi_{4,\rm{C}}(t) in Fig. 4 C captures the growth and suppression of correlated dynamics 8 ; 2 ; 39 . At low temperatures, pronounced peaks indicate long-lived collective motion within the ordered sublattice. As the system approaches T_{f} , these peaks sharpen, reflecting enhanced spatial correlation of particle motion. Below T_{f} , \chi_{4,\rm{C}}(t) broadens further, signifying the coexistence of mobile and immobile regions. In contrast, as the temperature rises toward the melting regime, the peak amplitude of \chi_{4,\rm{C}} drops abruptly, revealing the loss of persistent correlations as the sublattice destabilizes. This sudden suppression marks the transition from heterogeneous to homogeneous relaxation. Hence, the nonmonotonic evolution of \chi_{4,\rm{C}}(t) —growth near freezing followed by a sharp decline at melting—serves as a microscopic signature of the sublattice melting transition. To probe the connection between microscopic dynamics and transport, we examined the inverse relaxation times, 1/\tau_{\alpha,\mathrm{C}} and 1/\tau_{\chi_{4,\rm{C}}} , as a function of inverse temperature 1/T [Fig. 4 D ] (see the Materials and Methods for the definition of these time scales). Both quantities exhibit nearly Arrhenius behavior at high temperatures, but deviate sharply near T_{f} , indicating the onset of heterogeneous dynamics consistent with the MSD, F_{s,\rm{C}}(q,t) , and \chi_{4,\rm{C}}(t) results. In the Arrhenius regime, relaxation is governed by an approximately fixed activation barrier E_{a} , such that the characteristic relaxation time or diffusivity follows an Arrhenius law. This implies that particles cross comparable energy barriers irrespective of the instantaneous configuration of their neighbors, a hallmark of weakly correlated dynamics typical of normal liquids or dilute systems. However, near the sublattice melting transition, particle motion becomes increasingly cooperative. The system develops pronounced dynamical heterogeneity, where regions of mobile and immobile particles coexist, and structural relaxation requires collective rearrangements rather than isolated hopping events. As a result, the effective activation energy for relaxation, E_{\mathrm{eff}}(T) , becomes temperature-dependent and generally increases upon cooling because larger clusters of particles must move collectively. This leads to a super-Arrhenius temperature dependence, \tau_{\alpha,\mathrm{C}}\propto\exp\!\left[E_{\mathrm{eff}}(T)/k_{\mathrm{B}}T\right] , where E_{\mathrm{eff}}(T) increases as T decreases. The deviation from a linear behavior in an Arrhenius plot of inverse \tau_{\alpha,\mathrm{C}} versus 1/T therefore reflects the onset of collective dynamics. By contrast, purely local, single-particle hopping dynamics 31 would preserve a nearly constant activation barrier and thus Arrhenius behavior even near the transition. Accordingly, the emergence of correlated motion naturally implies a breakdown of the Stokes-Einstein (SE) relation. The corresponding Stokes-Einstein ratios D_{\rm C}\tau_{\alpha,\mathrm{C}} and D_{\rm C}\tau_{\chi_{4,\rm C}} (Fig. 4 E ) remain nearly constant at high T , show a modest reduction near T_{l} , and increase sharply upon approaching T_{f} . This violation of the Stokes-Einstein relation is a hallmark of growing dynamical heterogeneity: structural relaxation slows markedly as correlated regions develop, whereas diffusion remains comparatively fast due to cooperative carrier motion. The resulting decoupling between diffusion and relaxation time scales becomes pronounced once heterogeneous dynamics are long-lived 34 , signaling the onset of sublattice melting. While these signatures establish the presence of dynamical heterogeneity 30 ; 31 from a macroscopic transport perspective, they do not resolve its microscopic origin. To directly probe the spatial and temporal structure of heterogeneous carrier motion near T_{f} , we therefore examine particle-resolved displacements and local mobility patterns in Fig. 5 . To complement the dynamical observables discussed above, Figs. 5 A-B presents direct evidence of how structural and dynamical heterogeneity emerge near the sublattice-melting transition. Fig. 5 A maps the spatial distribution of particle mobility through \sqrt{\Delta r^{2}} over the interval [t_{0},t_{0}+\tau_{\alpha}] at T=0.45 , further enhanced by displacement-vector arrows whose length and colour encode the magnitude of particle motion. These arrows reveal not only the coexistence of liquid-like and immobile domains, but also the presence of concerted hopping events, manifested as string-like cooperative motion and correlated intermittent jumps within the mobile regions. Such collective displacements indicate that ion transport proceeds via cooperative rearrangements rather than independent single-particle hopping. Notably, enhanced mobility is also observed near the edges of crystallized domains, where a subset of particles transiently occupies interstitial positions, reflecting pronounced local fluctuations of the carrier environment. In contrast, the crystallized (red coloured) regions predominantly display short, irregular, and spatially heterogeneous displacement vectors, indicative of confined yet nonuniform vibrational dynamics. The corresponding static snapshot in Fig. 5 B , coloured by the local carrier-carrier nearest-neighbour averaged distance, exhibits a strikingly similar spatial pattern: particles in locally compressed, liquid-like environments (peach coloured) are spatially collocated with the dynamically mobile regions identified in Fig. 5 A , whereas particles in more expanded, ordered environments (navy blue coloured) coincide with dynamically frozen domains. Such a correspondence between structural softness and enhanced mobility closely parallels observations in glass-forming systems, where regions of local densification serve as loci of dynamic heterogeneity 30 ; 52 . The emergence of such spatial heterogeneity is consistent with the breakdown of carrier hyperuniformity upon approaching T_{m} , reflecting enhanced long-wavelength density fluctuations (see Supporting Information) 54 . Notably, the peach-colored regions correspond to carrier particles transiently occupying interstitial positions within the host lattice. As previously proposed for superionic AgI, transient occupation of interstitial sites leads to strongly anharmonic vibrational motion rather than simple harmonic oscillations about lattice sites 22 . The present observations therefore suggest that dynamic heterogeneity near sublattice melting is rooted in spatial variations of lattice softness and vibrational stability arising from interstitial carrier configurations. This directly motivates a quantitative analysis of lattice anharmonicity as the system approaches the sublattice-melting regime, which we pursue in the next section (Fig. 6 ). II.3 Growth of anharmonicity near the sublattice melting point Figure 6: Lindemann index and effect of anharmonicity: ( A–B )Time evolution of the Lindemann factor for carriers and hosts at packing fraction \varphi=0.85 . For the host lattice ( A ), the Lindemann factor shows a clear signature of lattice melting, while for the carriers ( B ) the growth is purely driven by anharmonic fluctuations rather than a melting transition. ( C ) Distinct anharmonic behavior is evident in the relative Debye-Waller factor: the host exhibits a linear dependence (black dotted line), indicating harmonic cage vibrations, whereas the carrier shows clear anharmonic effects (red solid line) with two distinct slopes, reflecting the increasing significance of nonlinear contributions near the onset of sublattice melting. To quantify the anharmonic vibrational state revealed above, we now turn to direct measures of lattice stability. We characterize the vibrational behavior of the system using the Lindemann index and the Debye-Waller factor, which capture deviations from harmonic motion and the progressive softening of the lattice. Then examine how variations in density regulate both the onset and the sharpness of sublattice melting. The resulting trends and their physical implications are discussed in the following sections. The evolution of the Lindemann index and the Debye-Waller factor (see Materials and Methods) reveals a pronounced enhancement of anharmonicity as the system approaches the sublattice melting point (Fig. 6 ). Figs. 6 A,B show the time evolution of the Lindemann index, \gamma_{{\rm L},\alpha}(t) , for the host and carrier particles at \varphi=0.85 . For the host lattice, \gamma_{{\rm L},{\rm H}}(t) exhibits a well-defined plateau reflecting harmonic vibrations within a stable crystal, followed by a rapid increase once \gamma_{{\rm L},{\rm H}}(t) exceeds the critical value \gamma_{{\rm L},{\rm H}}^{\rm c}\simeq 0.048 , consistent with the conventional Lindemann criterion for lattice melting and close to previously reported critical values \gamma_{\rm L}^{\rm c}\simeq 0.03 for crystalline solids 64 . In contrast, \gamma_{{\rm L},{\rm C}}(t) for the carriers increases smoothly without a sharp transition, indicating that their fluctuations are governed by intrinsic anharmonic vibrations rather than by a structural instability. The Debye-Waller factor, u_{\alpha}^{2} (see Materials and Methods), further quantifies this contrast in vibrational behavior [Fig. 6 C ]. For the host particles, u_{\rm H}^{2} increases linearly with temperature, u_{\rm H}^{2}\propto T , indicating that their dynamics are dominated by harmonic cage vibrations, as expected for an elastically stable lattice 33 . For the carriers, by contrast, u_{\rm C}^{2} exhibits a nonlinear temperature dependence of the form u_{\rm C}^{2}=aT+bT^{2} 33 , with the quadratic term bT^{2} becoming increasingly pronounced near the onset of sublattice melting. Within standard statistical-mechanical descriptions of lattice vibrations, such nonlinear temperature dependence arises from anharmonic terms in the effective potential, signaling local softening and strongly non-harmonic carrier fluctuations even while the host lattice remains elastically stable. Together, these results demonstrate that anharmonicity increases continuously as the sublattice melting point is approached: the host retains harmonic, lattice-governed vibrations, whereas the carrier particles display pronounced nonlinear responses, reflecting the breakdown of harmonic confinement and the onset of dynamically heterogeneous motion. II.4 Controlling the sublattice melting Figure 7: Density dependence of the diffusivity: ( A ) Temperature dependence of diffusivity of carriers ( D_{\mathrm{C}} ) as a function of inverse temperature ( 1/T ) for different area packing fractions. The results demonstrate that tuning the packing density effectively controls the onset of sublattice melting. With decreasing packing fraction, the transition becomes smoother, and distinct slope changes are observed across the five packing cases, indicating different dynamical regimes. Panels ( B–E ) correspond to \varphi=0.65 at decreasing temperatures ( T=0.500,0.250,0.150,0.100 ), showing progressive localization of carriers upon cooling. Panels ( F–I ) show configurations at T=0.500 (follow the black dashed line in A ) for increasing densities ( \varphi=0.70-0.85 ), demonstrating how higher packing suppresses anharmonic carrier motion and stabilizes the ordered sublattice. We next examine how the onset of sublattice melting varies systematically with particle density (Fig. 7 ). The temperature dependence of the carrier diffusivity, D_{\mathrm{C}} , for different packing fractions reveals that the melting behavior can be effectively tuned by density. As the packing fraction decreases, the transition becomes progressively broader, and the crossover region between solid-like and fluid-like states extends over a wider temperature range. This behavior indicates that the structural coherence of the host lattice weakens continuously as the density is reduced. Consistent with this trend, we further show that the carrier particles become increasingly anharmonic at lower densities (Figs. 7 A–I ). The broadening of the sublattice-melting regime thus directly correlates with the enhancement of carrier anharmonicity, demonstrating that these two phenomena are intrinsically linked. In particular, the density controls both the extent and sharpness of the transition, underscoring its pivotal role in governing the sublattice-melting process. Notably, even at the lowest examined density ( \varphi=0.65 ), a pronounced sublattice-melting region persists (Figs. 7 B–E ), with the carriers exhibiting liquid-like behavior even near room temperature. These findings provide fundamental insight into how sublattice melting can be deliberately induced or stabilized—potentially even at room temperature—through precise control of the packing density. Summary We have investigated a minimal two-dimensional binary mixture model of a superionic conductor to disentangle three intertwined ingredients of fast ion transport— sublattice melting , anharmonicity , and dynamic heterogeneity —and to clarify how density governs their emergence. The model captures selective sublattice melting within a chemically agnostic, coarse-grained framework. Host (H) particles interact via short-range steric repulsion, forming a rigid lattice, whereas carrier (C) particles interact through much softer, long-range Wigner-type forces that promote collective delocalization. This asymmetry in interaction range naturally produces distinct melting temperatures for the two sublattices, enabling a selective carrier-melting phase where the host remains ordered while the carrier sublattice becomes fluidlike (Fig. 1 ). Brownian dynamics was used for equilibration, followed by long microcanonical molecular-dynamics simulations to capture intrinsic dynamics (see Materials and Methods). Snapshots and radial distribution functions reveal clear selective disordering : the carrier sublattice loses positional order while the host remains crystalline (Figs. 1 , 2 ). Diffusivity-temperature plots identify three regimes (Fig. 2 A ): (I) crystalline, (II) a broad sublattice-melting region with mobile carriers in an ordered host, and (III) full melting. Carrier diffusivity evolves smoothly across regime II, whereas the host shows a sharp discontinuity, consistent with a first-order transition (Fig. 2 B,C ). Carrier trajectories (Fig. 3 ) visualize these regimes. At high T , motion is homogeneous and isotropic. Within regime II, carriers move cooperatively along host-lattice-aligned pathways forming transient honeycomb-like networks, while at lower T they exhibit heterogeneous motion, alternating between localized and collective migration. These patterns confirm that sublattice melting arises from correlated, anharmonic motion rather than uncorrelated hopping. Time-correlation analyses quantify these dynamics (Fig. 4 ). The mean-squared displacement and self-intermediate scattering function reveal transient caging and stretched-exponential relaxation, marking cooperative slow dynamics. The four-point dynamic susceptibility \chi_{4,\alpha}(t) peaks near the freezing side of regime II, indicating growing spatial correlations, and collapses in the liquid phase, signifying a crossover from heterogeneous to homogeneous relaxation. The Stokes-Einstein ratios for carriers D_{\mathrm{C}}\tau_{\alpha,\mathrm{C}} and D_{\mathrm{C}}\tau_{\chi_{4}} deviate strongly near the transition, confirming a breakdown of single-particle transport and the emergence of collective, anharmonic dynamics. Spatially resolved mobility maps reveal pronounced dynamic heterogeneity near the sublattice-melting transition (Fig. 5 A ). Carriers segregate into coexisting mobile and immobile domains, with string-like cooperative displacements and correlated hopping events indicating collective, rather than single-particle, transport. Enhanced mobility preferentially appears near the boundaries of crystallized regions, where particles transiently occupy interstitial positions. The corresponding static snapshots (Fig. 5 B ) demonstrate a strong correspondence between local structural softness and mobility: dynamically active regions coincide with locally compressed, liquid-like environments, while dynamically frozen regions remain structurally ordered. This spatial correlation parallels dynamic heterogeneity in glass-forming systems and implicates anharmonic lattice distortions as the microscopic origin of heterogeneous carrier dynamics near sublattice melting. The Lindemann index and Debye-Waller factor link these dynamical changes to local vibrations (Fig. 6 ). For hosts, u_{\rm H}^{2}\propto T , consistent with harmonic motion; for carriers, u_{\rm C}^{2}=aT+bT^{2} , where the quadratic term dominates near sublattice melting, evidencing enhanced anharmonicity even in an elastically stable host. Finally, varying the packing density tunes the onset and sharpness of sublattice melting (Fig. 7 ). Lower density broadens regime II and strengthens carrier anharmonicity, demonstrating that density controls both the range and cooperativity of the transition by reshaping the local energy landscape. Together, these results provide a unified microscopic picture linking order-disorder dynamics, anharmonic vibrations, and dynamical heterogeneity. Unlike conventional mean-field or single-particle hopping descriptions, which assume independent ion motion in a fixed lattice potential, our model reveals that fast ionic transport emerges from collective, anharmonic rearrangements constrained by the crystalline host. This cooperative mechanism naturally couples local vibrational softening to large-scale sublattice disorder, bridging the gap between microscopic lattice dynamics and macroscopic transport. These insights suggest that sublattice melting can be engineered—potentially even at ambient conditions—via density control , offering design principles for robust, high-conductivity solid electrolytes. The roles of dimensionality and carrier size in shaping transport behavior are examined in detail in the Supporting Information. Discussions While the present model captures the essential physics of sublattice melting, several directions remain for future exploration. To rigorously validate the microscopic mechanism proposed here, it is crucial to connect this coarse-grained framework with first-principles descriptions. Ab initio molecular dynamics and lattice-dynamical analyses could quantify anharmonic phonon modes, identify soft vibrational branches, and map potential-energy landscapes in representative ionic conductors such as AgI, PbF 2 , and Na 3 PS 4 . Comparing the temperature dependence of the Debye-Waller factors and Lindemann indices obtained from first-principles simulations with those derived in this study would establish a direct correspondence between lattice-level anharmonicity and emergent sublattice disorder. Furthermore, electronic-structure analyses could elucidate how bonding characteristics, ionic radii, and mass asymmetry govern the selective instability of carrier sublattices. Experimentally, our findings offer clear guidelines for materials design. In real systems, the effective density—or equivalently, the lattice volume or chemical pressure—can be tuned through compositional substitution, strain, or external pressure. We predict that reducing the effective density broadens the sublattice-melting regime and enhances carrier anharmonicity, implying that low-density or expanded-lattice phases may exhibit sublattice melting near room temperature. Advanced X-ray or neutron scattering measurements of the Debye-Waller factor could directly test the predicted crossover from harmonic to anharmonic behavior, while Raman and quasielastic neutron spectroscopy could probe phonon softening and correlated ionic motion for quantitative comparison with simulation. In conclusion, this study provides a quantitative and conceptual framework that bridges phenomenological modeling with an atomistic understanding of sublattice melting. Unlike conventional mean-field or single-ion hopping descriptions—which treat ionic transport as independent motion in a static potential—our results reveal that fast ion conduction arises from collective, anharmonic, and density-dependent dynamics constrained by the crystalline host. This cooperative mechanism unifies order-disorder transitions, vibrational anharmonicity, and dynamical heterogeneity within a single physical picture, offering a microscopic foundation for interpreting nonlinear transport behaviors observed in superionic conductors. Combining first-principles simulations with controlled experiments will be essential to verify the universality of these mechanisms and translate them into design principles for next-generation, room-temperature solid electrolytes. III Methods Molecular dynamics simulation codes were developed in C++, and post-processing, analysis, and figure generation were carried out using Python scripts with Matplotlib 24 and OVITO 50 . III.1 Simulations To reproduce the essential physics of sublattice melting within a minimal and chemically agnostic framework, we impose distinct interaction types on the two species. The host particles interact through short-range steric repulsion, forming a mechanically rigid crystalline framework, whereas the carrier particles experience a much softer, effectively long-range Wigner-type interaction that promotes collective delocalization within the host lattice. This asymmetry in interaction range and stiffness naturally yields widely separated melting temperatures for the two sublattices, enabling the emergence of a selective carrier-melting phase in which the carrier sublattice becomes fluidlike while the host lattice remains ordered. To place our approach in context, prior studies of superionic conductors have shown that ion transport and sublattice-selective melting can emerge from relatively simple interaction schemes. In tunnel-type solids, a balance of Coulombic, polarization, and short-range repulsive forces governs the off-axis migration pathways and activation barriers of mobile ions, producing size-selective mobility 14 . Likewise, it has been demonstrated that the superionic behaviour of \alpha -AgI can be captured using only Coulombic and soft-core repulsive interactions, with an ordered anion sublattice coexisting alongside a highly delocalized cation sublattice 15 . These works highlight that coarse-grained molecular dynamics with simplified but physically motivated interactions can reproduce essential features of ion-conducting phases. Our model follows this spirit but adopts an even more generalized, chemically agnostic formulation, retaining the core interplay between host confinement and carrier delocalization while avoiding system-specific assumptions. Table 1: Model parameters used in the simulations. N_{\mathrm{C}}{:}N_{\mathrm{H}} m_{\mathrm{C}} m_{\mathrm{H}} \epsilon_{\mathrm{CC}} \epsilon_{\mathrm{HH}} \epsilon_{\mathrm{CH}} \sigma_{\mathrm{C}} \sigma_{\mathrm{H}} \sigma_{\mathrm{CH}} 1:1 1.0 1.0 0.001 1.0 1.0 0.154 1.0 0.577 To this end, we consider a two-dimensional binary mixture comprising a total of N=512 particles, with equal numbers of host ( N_{\mathrm{H}}=256 ) and carrier ( N_{\mathrm{C}}=256 ) species. As shown in Fig. 1 , the initial configuration of the binary mixture is constructed in a hexagonal packing arrangement, where the host particles define the underlying lattice and the carriers are distributed within the interstitial sites. The interparticle interactions are modeled using the Weeks-Chandler-Andersen (WCA) potential, a truncated and shifted form of the Lennard-Jones potential, as detailed in the following. For a pair of particles i and j separated by a distance r_{ij}=|\mathbf{r}_{i}-\mathbf{r}_{j}| , the interaction potential is given by \displaystyle U \displaystyle=\sum_{\langle i,j\rangle}U_{ij}(r_{ij}), (1) \displaystyle U_{ij}(r_{ij}) \displaystyle=\begin{cases}4\epsilon_{ij}\left[\left(\dfrac{\sigma_{ij}}{r_{ij}}\right)^{12}-\left(\dfrac{\sigma_{ij}}{r_{ij}}\right)^{6}\right]+\epsilon_{ij},&r_{ij}\leq r_{c},\\[8.0pt] 0,&r_{ij}>r_{c}.\end{cases} where \sigma_{ij} is the effective particle diameter for the interacting pair and \epsilon_{ij} sets the energy scale. The cutoff distance r_{c} = 2^{1/6}\sigma_{ij} ensures that the potential is purely repulsive. In our simulation, we use \sigma_{\mathrm{H}} , \epsilon_{\mathrm{HH}} , m_{\mathrm{H}} , and \sqrt{m_{\mathrm{H}}\sigma_{\mathrm{H}}^{2}/\epsilon_{\mathrm{HH}}} as units of length, energy, mass, and time, respectively. The interaction parameters (\sigma_{ij},\epsilon_{ij}) for all pairs of particles are summarized in Table 1 . See Fig. 1 for detailed visualization. The host-host ( \mathrm{HH} ) and carrier-host ( \mathrm{CH} ) interactions are defined with a common energy scale \epsilon_{ij}=1.0 , whereas the carrier-carrier ( \mathrm{CC} ) interaction is deliberately chosen to be much weaker, \epsilon_{\mathrm{CC}}=10^{-3} . This parametrization effectively suppresses direct mutual exclusion between carriers and instead promotes collective, long-range-mediated dynamics within the confining host lattice. The physical motivation for this choice follows earlier modeling strategies in which mobile ions interact through weak, extended interactions that encode long-range correlations rather than explicit short-range repulsion 15 ; 41 . In contrast to those works, however, we do not introduce explicit charge-specific Coulomb interactions. Instead, we employ a non-additive, long-range interaction in a minimal form , designed to capture the essential collective effects of carrier motion while remaining chemically agnostic. Remarkably, despite this simplification, the model robustly reproduces key phenomenology associated with superionic behavior, including selective sublattice melting and strongly heterogeneous carrier dynamics. This demonstrates that explicit electrostatics are not a prerequisite for sublattice melting, which instead emerges as a collective phenomenon governed by lattice softness and long-range correlations among mobile carriers. Table 2: Variation of simulation box parameters with area packing fraction \varphi . \varphi a_{\mathrm{L\text{-}space}} \sigma_{\mathrm{CC}} L_{x} L_{y} 0.85 1.0440 1.0440 16.7040 14.4661 0.80 1.0773 1.0773 17.2362 14.9270 0.75 1.1126 1.1126 17.8016 15.4166 0.70 1.1516 1.1516 18.4256 15.9570 0.65 1.1951 1.1951 19.1216 16.5598 The lattice spacing a_{\mathrm{L\text{-}space}} , effective carrier size \sigma_{\mathrm{CC}} , and the corresponding box dimensions (L_{x},L_{y}) for the two-dimensional system at different area packing fractions \varphi are summarized in Table 2 . We employed both underdamped Brownian dynamics (BD) and Newtonian molecular dynamics (MD) to evolve the binary mixture. During the initial equilibration stage, the system was evolved under underdamped Langevin dynamics, \displaystyle m_{i}\frac{{\rm d}\mathbf{v}_{i}}{{\rm d}t}=-\zeta\mathbf{v}_{i}-\bm{\nabla}_{i}U+\bm{\eta}_{i}(t), (2) where \zeta is the damping coefficient and \bm{\eta}_{i}(t) is a Gaussian white noise term with zero mean: \langle\bm{\eta}_{i}(t)\rangle=\mathbf{0} and the variance \langle\bm{\eta}_{i}(t)\otimes\bm{\eta}_{j}(t^{\prime})\rangle=2k_{\mathrm{B}}T\zeta\,\delta_{ij}\delta(t-t^{\prime})\mathbf{1} due to fluctuation dissipation theorem and \mathbf{1} denotes the unit matrix. BD simulations were propagated for 2\times 10^{6} steps with timestep \Delta t_{\mathrm{BD}}=0.001 , corresponding to a total equilibration time t_{\mathrm{BD}}=2\times 10^{3} . After equilibration, the system was switched to microcanonical molecular dynamics (NVE ensemble) to follow the Hamiltonian evolution. MD simulations were performed for the 1\times 10^{8} steps with the time step \Delta t_{\mathrm{MD}}=0.0001 , corresponding to a total production run length t_{\mathrm{MD}}=1\times 10^{4} . In both stages, the system of N particles was propagated for sufficiently long times to ensure equilibration and statistical averaging. Brownian dynamics ensured relaxation into equilibrium configurations, whereas subsequent NVE molecular dynamics enabled the study of intrinsic dynamical processes without thermostatting. Finally, to assess the quantitative validity of the minimal model beyond qualitative phenomenology, we directly compared the carrier diffusivity obtained from simulations at packing fraction \varphi=0.70 with experimental diffusion data for \beta -alumina 27 in the Arrhenius regime associated with sublattice melting (see Supplementary Information for more details). By collapsing the Arrhenius portions of the simulated and experimental diffusivity curves, we extract an effective activation scale \varepsilon/k_{B}=1264.3~\mathrm{K} and a transport prefactor a^{2}/\tau_{0}=2.05\times 10^{-5}~\mathrm{cm}^{2}\,\mathrm{s}^{-1} . Details of the fitting procedure, scaling, and physical interpretation of these parameters are provided in the Supplementary Information. III.2 Radial distribution function The pair distribution function (PDF) of particles of type \alpha\in\{{\rm C,H}\} (carrier or host) is defined 19 as \displaystyle g_{\alpha\alpha}(\mathbf{r})=\frac{1}{\rho_{\alpha}N_{\alpha}}\left\langle\sum_{j\neq k,\;j,k\in\alpha}\delta\!\left(\mathbf{r}-\mathbf{r}_{j}+\mathbf{r}_{k}\right)\right\rangle, (3) where \rho_{\alpha}=N_{\alpha}/(L_{x}L_{y}) is the number density of particle of species \alpha and N_{\alpha} is the number of particles of species \alpha . The radial distribution function (RDF), g_{\alpha\alpha}(r) , was obtained by averaging the PDF over all angular directions. This function quantifies the local structural correlations within each particle species and provides a measure of the characteristic interparticle spacing and short/long-range order. III.3 Mean squared displacement The mean squared displacement (MSD) of particles of type \alpha\in\{{\rm C,H}\} (carrier or host) was computed as 36 \displaystyle\langle\Delta\mathbf{r}_{\alpha}(t)^{2}\rangle=\frac{1}{N_{\alpha}}\sum_{j\in\alpha}\left\langle\big[\mathbf{r}_{j}(t+t_{0})-\mathbf{r}_{j}(t_{0})\big]^{2}\right\rangle. (4) At long times, the MSD exhibits diffusive behavior, from which the self-diffusion coefficient D_{\alpha} is determined as \displaystyle\langle\Delta\mathbf{r}_{\alpha}(t)^{2}\rangle\sim 4D_{\alpha}t. (5) This analysis was conducted separately for the carrier and host species, allowing a direct comparison of their respective mobilities and of the temperature dependence of their transport dynamics. III.4 Self-intermediate scattering function The self-intermediate scattering function of particles of type \rm{C} (carrier) was calculated as 35 ; 19 \displaystyle F_{s,\mathrm{C}}(q,t)=\frac{1}{N_{\mathrm{C}}}\sum_{j\in\mathrm{C}}\left\langle\exp\!\left[i\mathbf{q}\cdot\big(\mathbf{r}_{j}(t+t_{0})-\mathbf{r}_{j}(t_{0})\big)\right]\right\rangle, (6) where \mathbf{q} is the wave vector, whose magnitude was chosen to correspond to the characteristic length scale of the host particles, i.e., near the first peak of the static structure factor. Relaxation dynamics was characterized by monitoring the temporal decay of F_{s,\mathrm{C}}(q,t) . The structural relaxation time \tau_{\alpha,\mathrm{C}} was determined by fitting F_{s,\mathrm{C}}(q,t) to the Kohlrausch-Williams-Watts (KWW) function, \displaystyle F_{s,\mathrm{C}}(q,t)\sim A\exp\{-[t/\tau_{\alpha,\mathrm{C}}]^{\beta}\}, (7) where A and \beta are the fitting parameters that represent the amplitude and stretching exponent, respectively. This analysis was performed independently for the carrier and host species, allowing a direct comparison of their microscopic relaxation behaviors and the temperature dependence of their dynamical responses. III.5 Four-point dynamic susceptibility To quantify dynamic heterogeneity consistent with our analysis of F_{s,\alpha}(q,t) , we evaluated the four-point dynamic susceptibility 39 ; 31 ; 47 for particles of type \alpha\in\{{\rm C,H}\} using the self-overlap order parameter. The single-particle overlap is defined as \displaystyle w^{(\alpha)}_{j}(t_{0}{+}t,t_{0})\equiv\Theta\!\Big(a-\big|\mathbf{r}_{j}(t_{0}{+}t)-\mathbf{r}_{j}(t_{0})\big|\Big), (8) where \Theta(\cdot) is the Heaviside step function and a is a microscopic cutoff that sets the cage scale. Unless otherwise noted, we set a=a_{\text{L-space}}/(2\sqrt{3}) , a geometric threshold corresponding to half the nearest-neighbor spacing in the host lattice, ensuring that the overlap function probes cage-breaking displacements on the host lattice length scale. The species-resolved (self) overlap is \displaystyle Q_{\alpha}(t)=\sum_{j\in\alpha}w^{(\alpha)}_{j}(t_{0}{+}t,t_{0}), (9) with averages taken over t_{0} (multiple time origins) and independent trajectories. The four-point dynamic susceptibility is then obtained from the variance of Q_{\alpha}(t) as \displaystyle\chi_{4,\alpha}(t)=\frac{V}{N_{\alpha}^{2}k_{\rm B}T}\Big(\langle Q_{\alpha}(t)^{2}\rangle-\langle Q_{\alpha}(t)\rangle^{2}\Big). (10) In reporting \chi_{4,\alpha}(t) , we highlight characteristic time: the peak time \tau_{\chi_{4}} that maximizes \chi_{4,\alpha}(t) . All quantities were computed separately for carrier and host particles to compare the species-resolved growth of dynamic correlations and their temperature dependence. III.6 Lindemann index To quantify species-resolved local relative fluctuations in particle positions, we computed the two-dimensional Lindemann index 1 ; 13 ; 17 ; 33 ; 42 ; 55 ; 59 \gamma_{{\rm L},\alpha}(t) for each component \alpha\in\{{\rm C,H}\} (carrier or host). Following the definition of the relative displacement between particles j and k , \displaystyle\mathbf{u}_{jk}(t)=\big[\mathbf{r}_{j}(t)-\mathbf{r}_{k}(t)\big]-\big[\mathbf{r}_{j}(0)-\mathbf{r}_{k}(0)\big], (11) the Lindemann index for each species is defined as \displaystyle\gamma_{{\rm L},\alpha}(t)=\frac{1}{N_{\alpha}}\sum_{j\in\alpha}\frac{1}{n_{j}}\sum_{k\in\alpha,~{\rm n.n.}(j)}\frac{\big\langle\big|\mathbf{u}_{jk}(t)\big|^{2}\big\rangle}{a_{\rm L-space}^{2}}. (12) Here, {\rm n.n.}(j) denotes the set of first-nearest-neighbor host particles surrounding particle j . The integer n_{j} is the coordination number of particle j , determined from the first minimum of the host-host radial distribution function. The Lindemann index \gamma_{{\rm L},\alpha}(t) thus measures the relative amplitude of the local vibrational motion of \alpha particles with respect to their nearest-neighbor distances on the host-particle length scale. This function was computed independently for carrier and host particles, allowing direct comparison of their local structural stability and the temperature dependence of their relative fluctuations. III.7 Debye-Waller factor The Debye–Waller factor quantifies the amplitude of vibrational motion in the caged regime and is closely related to the plateau value of the Lindemann index. For each species \alpha\in\{{\rm C},{\rm H}\} (carrier or host), we define \displaystyle u_{\alpha}^{2}=\gamma_{L,\alpha}(t_{\rm p}), (13) where \gamma_{L,\alpha}(t_{\rm p}) denotes the value taken at the characteristic turning time t_{\rm p} between the ballistic and diffusive regimes. This time marks the cage-confinement timescale at which both the mean-squared displacement and \gamma_{L,\alpha}(t) exhibit a transient plateau. For host particles ( \alpha={\rm H} ), u_{\rm H}^{2} corresponds directly to the plateau height of \gamma_{L,{\rm H}}(t) . For carrier particles ( \alpha={\rm C} ), the plateau is less pronounced due to stronger anharmonicity; accordingly, u_{\rm C}^{2} is evaluated at the same t_{\rm p} to represent their effective vibrational amplitude. This definition enables a consistent comparison of Debye–Waller factors between the two species and provides a microscopic proxy for the local stiffness of their respective environments. If u_{\alpha}^{2}\propto T holds, the dynamics in that regime can be regarded as effectively harmonic. III.8 Three-dimensional simulations To assess the robustness of the observed phenomenology against dimensionality, we performed complementary three-dimensional molecular dynamics simulations using a model constructed according to the same physical principles as the two-dimensional system. In particular, the interaction hierarchy, mass ratios, and relative size asymmetry between host and carrier species were kept identical to those employed in two dimensions, ensuring that any qualitative differences arise solely from dimensionality. The initial configuration was generated on a three-dimensional face-centered-cubic -like lattice composed of alternating host and carrier particles along all three spatial directions. The simulation box consisted of a periodic array of 4\times 4\times 4 unit cells, each containing an ordered arrangement of large host particles and smaller carrier particles, yielding a total particle number N\simeq 512 with equal populations of the two species. Host particles were assigned an effective diameter \sigma_{\mathrm{H}}=1.0 , while carrier particles were smaller, with \sigma_{\mathrm{C}}=0.154 , promoting preferential occupation of interstitial regions within the host framework. Both species were assigned equal masses. The lattice spacing was fixed at a_{\mathrm{L\text{-}space}}=1.5 , corresponding to a volume packing fraction of \varphi_{v}=0.6228 . The resulting cubic simulation box had a linear dimension L=L_{x}=L_{y}=L_{z}=4\times a_{\mathrm{L\text{-}space}}=6.0 , with periodic boundary conditions applied in all three spatial directions. This choice places the system in a mechanically stable crystalline regime for the host sublattice while allowing sufficient free volume for carrier delocalization. Interparticle interactions followed the same design as in two dimensions: host-host and host-carrier interactions were purely repulsive and sufficiently stiff to maintain long-range crystalline order, whereas carrier-carrier interactions were deliberately softened to suppress direct mutual exclusion and promote collective, correlated motion. The system was evolved using molecular dynamics with periodic boundary conditions in all three directions. After an initial equilibration stage, production runs were carried out in the microcanonical ensemble to probe intrinsic dynamical behavior without thermostat-induced artifacts. Consistent with the two-dimensional results, the three-dimensional simulations exhibit selective melting of the carrier sublattice (see Supplementary Information for more details). These findings suggest that the collective and anharmonic mechanisms identified here persist in fully three-dimensional systems and are not restricted to two dimensions. IV