Impact of ammonia hydrates and thermal gradients on the early geodynamic evolution of Enceladus’s south polar ice shell Authors: Katherine Villavicencio-Valero, Darío Mondavi-Sobby Published: 2026 ABSTRACT Enceladus, one of Saturn’s icy moons, exhibits active geology and compelling evidence for a subsurface ocean. The detection of ammonia in its plumes indicates that ammonia–water ice mixtures likely play a critical role in shaping its internal evolution by lowering the melting point of ice. However, the distinct thermodynamic properties of both ammonia monohydrate (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {NH}_{3} \cdot \hbox {H}_{2}\hbox {O}$$\end{document}) and ammonia dihydrate (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {NH}_{3} \cdot \hbox {2H}_{2}\hbox {O}$$\end{document}) have not been systematically evaluated in geophysical models. In this study, we present numerical geodynamic simulations that explore how each phase influences the thermal structure, onset of convection, and potential for thermodynamic conditions favorable for partial melting within Enceladus’s ice shell, under varying ice shell thicknesses, reference viscosity, and thermodynamic parameters. Our results show that both ammonia hydrates allow for the development of conditions conducive to localized melting and influence convective behavior under specific constraints. Additionally, the ammonia dihydrate system may sustain convective activity even at relatively high viscosities, allowing for the generation of localized partial melting. These findings highlight that ammonia hydrate phase composition might significantly influence geodynamic behavior and could contribute to conditions favorable for localized cryovolcanic resurfacing. Introduction Enceladus is one of the smallest moons in the Solar System, located in the brighter area of Saturn’s E ring1–3, with a radius of ∼ 250 km4,5. The South Pole is characterized as a region experiencing intense tectonic activity6–9, where water plumes eject material up to approximately 80 km above the surface. These plumes are composed of water vapour at temperatures ranging from 70 K to 150 K, and column density of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$1.5 \times 10^{16}~\text {cm}^{-2}$$\end{document}10; ice particles are emitted at rates between 150 kg s\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{-1}$$\end{document} and 300 kg s\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{-1}$$\end{document}10,11. Tidal stress may influence tectonic activity8,12, potentially acting as a heat source capable of melting subsurface materials such as the solid solution of ammonia dihydrate \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {NH}_{3}\cdot \hbox {2H}_{2}\hbox {O}$$\end{document}13,14. This internal heating mechanism could also contribute to the deformation of the ice crust, facilitating the expulsion of material through the water plumes that reach the surface15,16. According to Ref.17, the source of the plumes may be related to cryovolcanic activity, and the material released is most likely connected to the presence of hydrothermal activity at the seafloor of a liquid salty ocean layer beneath the icy crust2,6,17–20. The outer ice shell of Enceladus is estimated to have a mean thickness ranging between 10 km and 30 km21, with preferred values around 14 km – 26 km from libration data22 and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$23 \pm 4\ \text {km}$$\end{document} from gravity and shape analyses23–25. This shell is highly non-uniform, thinning to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$7 \pm 4\ \text {km}$$\end{document} at the South Pole, consistent with plume activity and enhanced tidal dissipation23, while other regions may be significantly thicker, reaching up to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sim 40\ \text {km}$$\end{document} in some regions26–28. The Ion and Neutral Mass Spectrometer (INMS) onboard the Cassini spacecraft detected ammonia within the plumes of Enceladus as a trace component, with concentrations not exceeding 0.5% in the plume composition11,20; it also exhibits a broad distribution throughout the outer Solar System, where it may occur in both hydrated and anhydrous forms11,13,29,30. Ammonia is also one of the most significant volatiles and mixtures condensing in the solar nebula and icy moons31–33. This compound significantly reduces the melting temperature of water ice by approximately 100 K34–36, enabling the formation of a subsurface ocean even under the cold conditions of Enceladus. This property facilitates internal differentiation and the development of compositional layers within the ice shell33,37. Early thermochemical models14 suggest that due to the relatively extreme condensation temperature of ammonia dihydrate \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {NH}_{3}\cdot \hbox {2H}_{2}\hbox {O}$$\end{document} in comparison with other volatiles, substantial quantities of ammonia (between 15 wt% to 30 wt% of the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {H}_{2}\hbox {O}$$\end{document} rich fraction) are dispersed within the icy bodies of the Solar System. A bulk ratio of about 0.15 is predicted for ammonia to water ice within the moons of Saturn and beyond the Solar System38,39. Furthermore, a solution of ammonia with less than 33 wt% \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {NH}_{3}$$\end{document} can stabilize liquid phases down to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sim 176\ \text {K}$$\end{document}, which is the peritectic point of the solution \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {NH}_{3}\cdot \hbox {2H}_{2}\hbox {O}$$\end{document}34,40,41. Although plume measurements indicate low concentrations, ammonia may be significantly enriched in the interior, where it participates in hydrate formation and influences thermodynamic behavior. Given its distinctive properties and its likely presence at concentrations above 1% in other icy worlds, ammonia may play a significant role in the geophysical and thermal evolution of icy bodies in the outer Solar System42. Numerical models based on ammonia hydrate spectra concluded that liquid water-ammonia cryomagma with particular concentrations of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {H}_{2}\hbox {O}$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {NH}_{3}$$\end{document}, such as the ammonia dihydrate \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {NH}_{3}\cdot \hbox {2H}_{2}\hbox {O}$$\end{document}, frequently overflows above the surface and cools down, removing \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {NH}_{3}$$\end{document} predominantly13. Enceladus is partly wrapped in Saturn’s magnetosphere, thus, ammonia and its constituents may be more frequently removed by sputtering under ion irradiation43. This likely results in significantly lower ammonia concentrations at the surface44–46 compared to those within the bulk from the interior. Additionally, ammonia may exhibit greater stability at the surface in the form of hydrates or as part of more complex compounds with other species47. Moreover, ammonia-rich environments may promote the segregation and concentration of low-melting-point phases within icy moons, as experimental studies indicate that the earliest melt in \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {NH}_{3}\cdot \hbox {H}_{2}\hbox {O}$$\end{document} systems is controlled by equilibrium involving ammonia dihydrate \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {NH}_{3}\cdot \hbox {2H}_{2}\hbox {O}$$\end{document}48. Laboratory studies on ammonia–bearing ices have shown that irradiation can alter their stability and abundance under surface conditions49. However, such processes are distinct from the bulk thermodynamic behavior governing melting in subsurface environments, which is controlled by pressure and composition–dependent phase relations. These melting temperatures are sufficiently low that radiogenic heating alone can raise internal temperatures to the solidus in icy moons with radii of only 500 km – 1000 km, potentially explaining the resurfacing observed on Saturnian and Uranian satellites within this size range50–52. Satellites with radii of a few hundred kilometers or less may be particularly relevant for \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {NH}_{3}\cdot \hbox {H}_{2}\hbox {O}$$\end{document} ices, as they are less likely to generate high internal temperatures than giant icy satellites48. Furthermore, developing accurate models of satellite interiors requires accounting for the pressure and temperature-dependent densities of ammonia hydrates, since these compounds may represent a substantial portion of the total ice mass. Unfortunately, many of the physical properties of ammonia hydrates and related liquids remain unknown. Published numerical models5,8,53–56 commonly assume that ammonia is incorporated into the water ice beneath Enceladus’ ice shell, playing a significant role in its melting and crystallization history. While previous studies have investigated the role of ammonia, they often do not quantify how different concentrations may affect the satellite’s thermal and structural evolution. In contrast, Ref.42 developed an evolutionary model of Enceladus that explicitly includes serpentinization processes and accounts for ammonia concentrations up to 7% by mass. Similarly, Ref.57 present a comparative evolution model incorporating ammonia contents of up to 10% alongside other volatiles. These models highlight the role of ammonia in long-term evolution, although they do not explicitly include the influence of tidal heating. Reference58 simulated the effects of tidal heating including the influence of ammonia in the South Pole of Enceladus, but it did not reproduce the consequences of melting during long scales of evolution. Reference59 investigated the response of varying the amount of ammonia within the internal evolution of Iapetus, finding out different influences in the creeps and internal differentiation leading to melting processes. Most existing models do not fully capture the thermophysical behavior of water ice below its melting point, as they primarily focus on conditions near the melting curve of pure \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {H}_{2}\hbox {O}$$\end{document} systems. This limitation emphasizes the necessity for advanced modeling approaches that incorporate subfreezing thermal conditions and explicitly consider the role of secondary volatiles like ammonia in modulating thermodynamic behavior. In this study, we develop numerical geodynamic simulations to investigate the thermal evolution of Enceladus’ south polar ice shell and the role of ammonia hydrates. The models isolate tidal heating as the sole internal energy source and evaluate the effects of ammonia monohydrate and dihydrate phases on melting conditions and convective behavior. Thermodynamic properties are parameterized using experimentally derived relationships, and melt generation is assessed using a postprocessing proxy. Rather than reproducing a unique present-day configuration, this work aims to identify physically plausible thermodynamic pathways and constrain the key parameters controlling heat transport, convection, and the onset of melting in this region. Results The ammonia dihydrate (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {NH}_{3}\cdot \hbox {2H}_{2}\hbox {O}$$\end{document}) case Figure 1 illustrates the convective structures, melt production, and volumetric tidal heating for simulations assuming a \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {NH}_{3}\cdot \hbox {2H}_{2}\hbox {O}$$\end{document} hydrate system. Experiments are presented for ice shell thicknesses of 50 km and 25 km, and for three reference viscosities: low (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{13}$$\end{document} Pa s), moderate (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{14}$$\end{document} Pa s), and high (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{15}$$\end{document} Pa s), arranged from top to bottom. The time (in Myr) corresponds to the point in each simulation when the system reaches a representative thermal and dynamic state. For both thicknesses, low viscosity models exhibit well-developed convective cells beneath a thin conductive lid (panels a and j). Despite this vigorous convective behavior, melt production remains negligible (panels b and k), and volumetric tidal heating is relatively low, in the order of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sim 3\times 10^{-7}$$\end{document} W m\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{-3}$$\end{document} - \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\ 4\times 10^{-7}$$\end{document} W m\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{-3}$$\end{document} (panels c and l).Fig. 1Snapshots of convective plumes, melt proxy, and volumetric tidal heating in simulations assuming a bulk ocean composition of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {NH}_{3}\cdot \hbox {2H}_{2}\hbox {O}$$\end{document}. The first three rows correspond to models with a 50 km ice shell thickness, while the lower three represent cases with a thickness of 25 km. Each panel indicates the reference viscosity, density, ice shell thickness, and the simulation time, either at the onset of partial melting or when convection has reached a quasi-stationary state.Fig. 2Snapshots of convective plumes, melt proxy, and volumetric tidal heating for simulations using a bulk ocean composition of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {NH}_{3}\cdot \hbox {H}_{2}\hbox {O}$$\end{document}. The first three rows correspond to an ice shell thickness of 50 km, while the lower three represent cases with a thickness of 25 km. Each model indicates the imposed reference viscosity, density, shell thickness, and the time required to reach the onset of partial melting; for simulations in which melting does not occur, the system was halted when the thermal state reached its maximum temperature. Snapshots of convective plumes, melt proxy, and volumetric tidal heating in simulations assuming a bulk ocean composition of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {NH}_{3}\cdot \hbox {2H}_{2}\hbox {O}$$\end{document}. The first three rows correspond to models with a 50 km ice shell thickness, while the lower three represent cases with a thickness of 25 km. Each panel indicates the reference viscosity, density, ice shell thickness, and the simulation time, either at the onset of partial melting or when convection has reached a quasi-stationary state. Snapshots of convective plumes, melt proxy, and volumetric tidal heating for simulations using a bulk ocean composition of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {NH}_{3}\cdot \hbox {H}_{2}\hbox {O}$$\end{document}. The first three rows correspond to an ice shell thickness of 50 km, while the lower three represent cases with a thickness of 25 km. Each model indicates the imposed reference viscosity, density, shell thickness, and the time required to reach the onset of partial melting; for simulations in which melting does not occur, the system was halted when the thermal state reached its maximum temperature. Moderate viscosity models display different thermal structures; in the 50 km ice shell, convective cells are well developed (panel d), whereas in the 25 km case, they appear more irregular and spatially heterogeneous near the model boundaries (panel m). Nonetheless, localized regions where temperatures exceed the solidus develop in both cases (panels e and n), supported by enhanced tidal dissipation of approximately \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sim 6 \times 10^{-7}\ \text {W m}^{-3}$$\end{document} (panels f and o). These conditions are reached earlier than in the low-viscosity cases for both ice shell thicknesses (panels e and n), originating in the deeper regions of the upwellings and persisting over longer timescales. High viscosity simulations show a sluggish formation of convective plumes in the 25 km (panel p), while it displays a thicker stagnant lid above the convective cells for the model of 50 km (panel g). In the 50 km model, partial melting occurs along the whole upwellings starting from the deeper regions (panels h), whereas no melt production is observed in the 25 km ice shell (panel q). Tidal dissipation remains lower than in moderate and low reference viscosity cases for the 50 km thickness experiments, while it does not exert a significant influence in the 25 km cases (panel r). The ammonia monohydrate (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {NH}_{3} \cdot \hbox {H}_{2}\hbox {O}$$\end{document}) case Figure 2 shows the results for the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {NH}_{3} \cdot \hbox {H}_{2}\hbox {O}$$\end{document} composition, using the same structure and organization as Fig. 1. For an ice shell thickness of 50 km, for both moderate and high reference viscosity models (panels e and h), localized regions where temperatures exceed the solidus are identified. Although both cases reach partial melting conditions, the influence of tidal dissipation differs; for the moderate reference viscosity it reaches values up to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$8 \times 10^{-7}\ \text {W m}^{-3}$$\end{document} (panel f), while the high viscosity experiment displays slightly lower values of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$5 \times 10^{-7}$$\end{document} to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$7 \times 10^{-7}\ \text {W m}^{-3}$$\end{document} (panel i). In addition, the convective structures feature thicker stagnant lids for the high reference viscosity case (panel g), in comparison to the moderate reference viscosity model (panel d). By contrast, the low viscosity model with 50 km thickness (panel a) displays well-developed convection beneath a relatively thin stagnant lid. This model does not reach melting conditions (panel b), and tidal dissipation remains minimal (panel c), indicating a possible correlation between tidal heating and melt production, similar to the outputs obtained for the same configuration in Fig. 1. Reducing the ice shell thickness to 25 km drastically limits melt production across all reference viscosity regimes. Although the low reference viscosity model continues to generate relatively higher tidal heating, the ice shell thickness likely inhibits the development of partial melting. These results suggest that the possible content of ammonia monohydrate, while it reduces the melting temperature of water ice, it is not self-sufficient to enable extensive melt generation under the influence of tidal dissipation alone. Instead, it might act as a thermal stabilizer, helping to maintain the integrity of the ice shell and supporting conditions that may prolong the persistence of a subsurface ocean. Discussion To better understand the physical mechanisms controlling the development and extent of partial melting, we analyze the geodynamic properties derived from the results. Figure 3 illustrates the correlation between ice shell thickness, reference viscosity, and temperature gradients for ammonia monohydrate (right column) and dihydrate (left column). Low and moderate reference viscosity models exhibit steeper thermal gradients and thinner stagnant lids for the 50 km ice shell thickness, promoting more efficient heat transport. In contrast, high viscosity experiments are marked by thicker conductive lids that inhibit basal warming, limiting the approach to the ammonia–water ice melting threshold, specifically for the 25 km models. Monohydrate models exhibit more confined thermal structures, particularly within 25 km ice shells, with further narrowing observed at higher reference viscosities. This could imply that, for the monohydrate phase, the onset of convection is more sensitive to perturbations in rheological properties, leading to more restricted heat transport than in the dihydrate mixture. However, dihydrate models demonstrate a slightly broader thermal window for convective transport, suggesting an enhanced ability to moderate the effects of localized tidal heating. Moderate and low reference viscosity provides insight into transitional regimes; for thinner ice shells, small changes in viscosity can shift the system from an active convective state to a sluggish regime with significant conductive dominance even though low reference viscosity did not produce partial melting as shown in Figs. 1 and 2. Moreover, in both \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {NH}_{3}\cdot \hbox {H}_{2}\hbox {O}$$\end{document}/50 km and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {NH}_{3}\cdot \hbox {2H}_{2}\hbox {O}$$\end{document}/25 km experiments, the overlap between the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{13}$$\end{document} Pa s and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{14}$$\end{document} Pa s curves may reflect a compensatory effect between viscosity and basal temperature. Although the reference viscosities differ by an order of magnitude, the corresponding basal temperatures, constrained by composition specific melting thresholds (194 K and 176 K, Table 1), lead to similar effective viscosities at the ice ocean interface. This thermal–rheological feedback reduces the expected differences in convective behavior, resulting in similar thermal profiles between the two cases. Figure 4 shows the profiles of vertical heat flux as a function of ice shell thickness. These profiles illustrate how thermal energy is transported from the base of the ice shell toward the surface through a combination of conductive and convective processes, with deviations from purely conductive trends reflecting the contribution of convection. Here, vertical heat flux increases with increasing reference viscosity and ice shell thickness, reflecting the reduced efficiency of convection under mechanically more resistant and thermally stable conditions. Within the 25 km ice shell, high reference viscosity yields low vertical heat flux values of approximately 0.01 W m\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{-2}$$\end{document}, indicating a regime dominated by conduction, regardless of chemical composition. Moderate reference viscosities produce moderately higher fluxes, reaching about 0.04 W m\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{-2}$$\end{document} for ammonia dihydrate simulations and 0.03 W m\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{-2}$$\end{document} for monohydrate models. In contrast, low reference viscosity results in elevated vertical heat flux values for the dihydrate case, accompanied by an anomalous profile that may reflect a reorganization of convective structures in response to strong thermal gradients. This behavior, occurring in the absence of melting conditions, suggests a dynamic yet stable convective regime characteristic of low-viscosity systems under limited tidal heating influence. In contrast, for 50 km ice shells, the vertical heat flux profiles become more complex, particularly at low to moderate reference viscosities. Variations along the profiles indicate the presence of thermally driven instabilities influencing heat transport. Conversely, high reference viscosity models exhibit nearly linear or weakly varying profiles, with values remaining below 0.02 W m\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{-2}$$\end{document}, consistent with a predominantly conductive regime.Fig. 3Relationship among ice shell thickness, temperature, and reference viscosity. Each panel illustrates the thermal structure of the modeled ice shell for a given reference viscosity regime, with data points color–mapped by temperature, symbol size scaled according to the reference viscosity value (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{13}$$\end{document} Pa s, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{14}$$\end{document} Pa s, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{15}$$\end{document} Pa s), and chemical composition respectively. Color bars indicate the temperature gradient across the shell. Relationship among ice shell thickness, temperature, and reference viscosity. Each panel illustrates the thermal structure of the modeled ice shell for a given reference viscosity regime, with data points color–mapped by temperature, symbol size scaled according to the reference viscosity value (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{13}$$\end{document} Pa s, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{14}$$\end{document} Pa s, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{15}$$\end{document} Pa s), and chemical composition respectively. Color bars indicate the temperature gradient across the shell. Although these simulated values are lower than the maximum heat fluxes reported by Cassini’s CIRS instrument60, estimated to reach up to 0.5 W m\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{-2}$$\end{document} over active regions of the South Pole, they remain consistent with the mean background flux of 0.1 W m\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{-2}$$\end{document} inferred for broader polar terrains. This difference is primarily attributed to the greater ice shell thickness adopted in our models (25 km – 50 km), compared to geophysical estimates suggesting thinner south polar shells of 5 km – 10 km. Thicker shells increase thermal resistance and reduce the efficiency of heat transport toward the surface. In terms of composition, ammonia-rich systems slightly enhance the efficiency of vertical heat transport within thicker shells, producing localized fluxes that, although still below the upper estimates from Cassini, fall within the lower observational range (0.1 W m\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{-2}$$\end{document} – 0.5 W m\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{-2}$$\end{document}). In thinner ice shells, this effect becomes less pronounced. The low reference viscosity monohydrate model exhibits reduced vertical heat flux near the base despite active convection, suggesting limited thermal amplification under conditions of weak tidal dissipation.Fig. 4Vertical heat flux profiles as a function of depth for different ice shell thicknesses in ammonia–water systems. Each curve corresponds to a reference viscosity of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{13}$$\end{document} Pa s, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{14}$$\end{document} Pa s, or \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{15}$$\end{document} Pa s, spanning a plausible rheological range for Enceladus’s ice shell under the modeled thermal conditions. Higher heat flux values indicate more efficient convective heat transport, whereas near-linear or weakly varying profiles are consistent with conduction-dominated regimes. Vertical heat flux profiles as a function of depth for different ice shell thicknesses in ammonia–water systems. Each curve corresponds to a reference viscosity of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{13}$$\end{document} Pa s, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{14}$$\end{document} Pa s, or \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{15}$$\end{document} Pa s, spanning a plausible rheological range for Enceladus’s ice shell under the modeled thermal conditions. Higher heat flux values indicate more efficient convective heat transport, whereas near-linear or weakly varying profiles are consistent with conduction-dominated regimes. Our results emphasize that vertical heat flux and its geophysical implications are modulated by the interplay between ice shell thickness, ammonia hydrate composition, and rheological constraints. While moderate reference viscosities in dihydrate systems favor the development of localized melting conditions, high viscosities and thinner ice shells tend to suppress these conditions, promoting conduction-dominated regimes. Enhanced vertical heat transport in localized regions may support sustained cryovolcanic activity and the long-term stability of subsurface liquid reservoirs beneath Enceladus’s ice shell. Although the simulated heat fluxes remain below the peak values inferred from CIRS observations, they fall within the lower observational bounds, indicating that the modeled processes are consistent with background thermal conditions. These results highlight the importance of considering the combined effects of tidal forcing amplitude, regional shell thickness variations, and phase-dependent rheology when interpreting surface heat flux anomalies detected by CIRS60. While vertical heat flux profiles provide insight into the balance between conductive and convective heat transport, they do not fully capture the internal dynamics of Enceladus’s ice shell. In particular, they do not directly quantify the timing or extent of melting conditions. To better constrain the persistence and spatial variability of subsurface liquid layers, it is therefore necessary to analyze the temporal evolution of partial melting conditions over short geological timescales. Figure 5 addresses these analyses by quantifying the temporal evolution of melting conditions within Enceladus’ ice shell, based on the computed melt fraction proxy (ϕ) and the corresponding mean temperature for the experiments displayed in Figs. 1 and 2. Each simulation explores a distinct combination of melt fraction proxy, reference viscosity, and ice shell thickness. The formulation for calculating the melt fraction proxy is described in Eq. (13). The model composed by \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {NH}_{3}\cdot \hbox {2H}_{2}\hbox {O}$$\end{document}a), produces elevated melt fraction values that surpass \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sim 1.2 \times 10^{-5}$$\end{document} within ∼ 2 Myr, accompanied by a temperature increase from ∼ 125 K to ∼ 170 K. This increase may indicate that both ice shell thickness and rheological conditions favor the onset of melting conditions when tidal dissipation is higher.Fig. 5Temporal profiles of mean temperature (dashed green) and melt fraction (solid blue) for NH\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$_3{\cdot }$$\end{document}2H\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$_2$$\end{document}O and NH\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$_3{\cdot }$$\end{document}H\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$_2$$\end{document}O mixtures, reference viscosities (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{14}$$\end{document} Pa s and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{15}$$\end{document} Pa s), and ice shell thicknesses of 25 km and 50 km. Melt fraction is normalized by following the approach described in Eq. (13). Temporal profiles of mean temperature (dashed green) and melt fraction (solid blue) for NH\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$_3{\cdot }$$\end{document}2H\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$_2$$\end{document}O and NH\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$_3{\cdot }$$\end{document}H\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$_2$$\end{document}O mixtures, reference viscosities (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{14}$$\end{document} Pa s and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{15}$$\end{document} Pa s), and ice shell thicknesses of 25 km and 50 km. Melt fraction is normalized by following the approach described in Eq. (13). When the ice shell thickness is 50 km with the same moderate reference viscosity (Fig. 5b), melting conditions are reached later compared to the experiment illustrated in Fig. 5a. The melt fraction proxy reaches a value of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sim 1.1 \times 10^{-5}$$\end{document}, while the temperature rises more gradually, eventually exceeding 200 K. This outcome underscores the competing roles of tidal dissipation and ice shell thickness in modulating heat retention by delaying the onset of melting conditions. In Fig. 5c, the melt fraction proxy barely exceeds \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sim 4.5 \times 10^{-6}$$\end{document} after 20 Myr, and the mean temperature remains constrained below ∼ 170 K. These results suggest that elevated viscosity suppresses tidal dissipation and restricts the mobility of partially molten material, ultimately reducing the extent of melting conditions even under sustained thermal gradients. In contrast, the monohydrate system (Fig. 5d) yields slightly lower melt fraction values of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sim 9 \times 10^{-6}$$\end{document} compared to previous cases, with temperatures approaching 200 K. Furthermore, Fig. 5e shows melt fraction proxy values reaching around \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sim 6 \times 10^{-7}$$\end{document} as the temperature gradually rises to ∼ 170 K. These experiments indicate that moderate reference viscosity promotes the development of melting conditions, whereas high viscosity inhibits their occurrence. However, it is important to note that the computed melt fraction proxy values remain extremely small (on the order of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{-7}$$\end{document} to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{-5}$$\end{document}) and do not represent significant bulk melting. Instead, they could reflect the onset of melting conditions and the thermodynamic potential for localized melt generation. This contrast provides important constraints on the conditions that may allow the long–term persistence of melting regimes within Enceladus’ ice shell. Overall, the results suggest that the interplay between ice shell thickness, ammonia composition, and rheological properties plays a fundamental role in controlling both the extent and persistence of internal melting under plausible geophysical conditions. To advance our understanding of the thermal and mechanical processes underlying the behaviors described above, it is essential to examine the material properties that govern energy transport and buoyancy within Enceladus’ ice shell. In particular, variations in thermal expansivity and density are key factors controlling the onset and intensity of convection, as well as the ability of the system to reach melting conditions. Figure 6 quantifies how ammonia hydrate composition influences thermal expansivity across all modeled scenarios, providing a thermodynamic perspective that complements the convective analysis. The figure shows the relationship between thermal expansivity and temperature for ammonia monohydrate and dihydrate systems, for both ice shell thicknesses (25 km and 50 km). Thermal expansivity is calculated using the empirical power-law formulation described in Eq. (7). In all models, thermal expansivity increases with temperature, reflecting enhanced buoyancy forces associated with thermal expansion. Moderate reference viscosity models consistently exhibit higher thermal expansivity values compared to both lower and higher viscosity cases, a pattern observed across compositions and ice shell thicknesses. This behavior likely reflects the fact that intermediate viscosity regimes reach temperatures sufficient to enhance thermal expansivity while still maintaining efficient heat redistribution. The 50 km ice shells systematically display slightly higher thermal expansivity values than the 25 km cases. These results indicate that thermal expansivity is influenced by the combined effects of reference viscosity, ammonia composition, and ice shell thickness. These variations have important implications for the underlying convective dynamics, because thermal expansivity directly controls buoyancy forces. Higher values therefore enhance the efficiency of convective heat transport. The peak thermal expansivity observed in the moderate reference viscosity models likely contributes to the development of the convective regimes identified in Figs. 1 and 2, and to the onset of melting conditions in those cases. Complementary density analyses (Supplementary Information) further support this interpretation. Even modest temperature variations produce density contrasts that enhance buoyancy forces, highlighting the role of ammonia hydrate mixtures in sustaining thermally active convective regimes under geophysically plausible internal conditions for Enceladus. Thermal expansivity thus emerges as a central parameter linking the thermodynamics of ammonia–water ice hydrates to their large–scale dynamics. It controls how efficiently temperature perturbations translate into density contrasts, such that even modest compositional or rheological variations can shift the system between a stagnant lid and an actively convecting regime within the ice shell. In the context of Enceladus, where surface tidal stresses and internal heating are coupled, a 10% – 20% increase in α enhances buoyancy-driven instabilities, potentially accelerating convective processes and expanding the thermal window over which melting conditions can be sustained. Consequently, regions that reach the elevated α values identified in the moderate reference viscosity models are more likely to sustain melting conditions, enhance tidal dissipation, and contribute to the south polar heat flux observed by Cassini. Recognizing the sensitivity of α to both temperature and hydrate phase not only clarifies the contrasting behaviors captured in Figs. 1 and 2, but also provides a predictive framework for identifying regions where cryovolcanic activity is most likely to persist within Enceladus’s heterogeneous ice shell. The results presented in this study suggest that ammonia hydrates act not only as antifreezing agents, but also as active modulators of ice shell dynamics. Through their influence on phase transitions, thermal expansivity, density stratification, and convective onset conditions, these compounds contribute to the persistence of melting conditions and influence the spatial distribution of tidal heating. This integrated framework improves our understanding of the thermophysical evolution of volatile-rich ice shells under realistic geophysical constraints and provides a basis for interpreting the geological and thermal signatures observed on Enceladus and other icy moons.Fig. 6Temperature dependence of the volumetric thermal expansion coefficient (α) for the two ammonia hydrate mixtures considered in this study: NH\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$_3{\cdot }$$\end{document}2H\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$_2$$\end{document}O and NH\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$_3{\cdot }$$\end{document}H\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$_2$$\end{document}O. Results are shown for two representative ice shell thicknesses (25 km and 50 km). In each panel, α is computed using the empirical power-law relation \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\alpha (T) = A T^B$$\end{document}, with coefficients A and B derived from Ref.41 and assumed as a first-order approximation for planetary ice shells. Colored lines and markers represent three reference viscosities (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{13}$$\end{document} Pa s, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{14}$$\end{document} Pa s, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{15}$$\end{document} Pa s) used in the coupled thermodynamic models to illustrate the sensitivity of thermal expansion to temperature and rheology. Temperature dependence of the volumetric thermal expansion coefficient (α) for the two ammonia hydrate mixtures considered in this study: NH\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$_3{\cdot }$$\end{document}2H\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$_2$$\end{document}O and NH\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$_3{\cdot }$$\end{document}H\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$_2$$\end{document}O. Results are shown for two representative ice shell thicknesses (25 km and 50 km). In each panel, α is computed using the empirical power-law relation \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\alpha (T) = A T^B$$\end{document}, with coefficients A and B derived from Ref.41 and assumed as a first-order approximation for planetary ice shells. Colored lines and markers represent three reference viscosities (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{13}$$\end{document} Pa s, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{14}$$\end{document} Pa s, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{15}$$\end{document} Pa s) used in the coupled thermodynamic models to illustrate the sensitivity of thermal expansion to temperature and rheology. Even though the ammonia concentrations adopted in our models correspond to an end-member scenario (∼33 wt%), this choice is intended to isolate the thermodynamic effects of ammonia on melting behavior under extreme conditions. These simulations are not designed to represent the present-day composition of Enceladus, but rather to quantify the maximum impact that ammonia may exert on the thermal structure and dynamics of the ice shell. In this context, the results demonstrate that ammonia hydrate phases can significantly reduce melting temperatures and facilitate the development of melting conditions. This suggests that even lower ammonia concentrations, more consistent with plume measurements, may still exert a measurable influence on the thermal and geodynamic evolution of the south polar region. Although the numerical experiments presented here provide new insights into the thermodynamic and convective behavior of ammonia–water ice shells, several limitations should be considered. The key thermodynamic properties, including thermal expansivity, density variation, and melting temperatures, are primarily derived from the experimental data of Ref.41. These datasets are constrained to near-surface pressures and laboratory environments; consequently, the empirical relationships used in this study are extrapolated beyond their original calibration range to approximate behavior under moderate hydrostatic pressures relevant to Enceladus’s interior. Another aspect of the models presented in this study is the exclusion of additional chemical species, such as salts and clathrate hydrates, which may further depress melting temperatures and modify the thermal conductivity and viscosity of the ice–ocean system. Similarly, the influence of grain size evolution, plastic yielding, and non–Newtonian rheologies is not explicitly included, although these factors can significantly affect ice shell deformation and heat transport. Lastly, tidal dissipation is imposed as a volumetric heating term based on prescribed viscosities and shell thickness, without fully coupling the tidal forcing to the evolving viscoelastic structure of the ice. This simplification may underrepresent spatial variations in heating rates, particularly in the context of complex orbital resonances and variable eccentricity. Nevertheless, our results provide a reasonable approximation of the influence of ammonia hydrate phases on the thermal evolution and the development of melting conditions within Enceladus’s ice shell. Conclusions This study indicates that the incorporation of ammonia monohydrate and dihydrate phases can significantly influence the geodynamic evolution of Enceladus’s ice shell. By systematically varying chemical composition, ice shell thickness, and reference viscosity, our simulations reveal how these hydrates affect the onset and persistence of melting conditions. Dihydrate systems tend to expand the range of conditions favorable for convective activity, while monohydrate regions show greater sensitivity to rheological and thickness variations. These results suggest that small-scale compositional heterogeneities may promote localized regions where melting conditions are reached, with contrasting thermal stability. Furthermore, our analysis highlights the central role of thermal expansivity in sustaining buoyancy–driven convection, even under moderate tidal heating. This thermodynamic control underscores the importance of accurately representing ammonia hydrate behavior in models of icy satellite interiors. Oversimplified parameterizations may lead to misinterpretations of the development and persistence of melting conditions, as well as the stability of subsurface liquid reservoirs. Beyond its geophysical implications, this study also has relevance for the broader search for habitable environments in the outer Solar System. The ability of ammonia hydrates to lower melting temperatures and sustain melting conditions under modest internal heating suggests that similar volatile–rich ice shells, such as those hypothesized for Uranian satellites or cold exoplanets, may remain thermally active over extended timescales. This thermodynamic buffering may help preserve subsurface liquid reservoirs, enabling chemical exchange and potentially supporting prebiotic conditions. Methods Geodynamic models setup The internal structure of icy satellites reflects a balance between composition, thermal evolution, and external forcing. In bodies such as Enceladus, this balance results in the coexistence of a cold outer ice shell and a warmer interior, where liquid reservoirs may persist under suitable thermodynamic conditions61,62. The presence of volatile compounds, particularly ammonia, can substantially alter this balance by modifying melting relations, density contrasts, and the efficiency of heat transport within the ice shell34,35. Rather than attempting to reproduce the full compositional complexity of Enceladus, we adopt a simplified framework that captures the dominant thermodynamic controls on ice shell dynamics. Specifically, we consider an idealized configuration consisting of an ice shell overlying a subsurface ocean, where thermal gradients, rheology, and ammonia hydrate phases govern the system evolution. The model incorporates several simplifying assumptions: the ice shell is treated as a homogeneous continuum with temperature–dependent viscosity; compositional variations are represented through ammonia hydrate end–members assumed to be part of the bulk ocean composition; and tidal dissipation is imposed as a spatially uniform volumetric heating term that is not dynamically coupled to viscoelastic evolution. Radiogenic heating, grain-size evolution, and non-Newtonian rheologies are neglected. This framework enables a systematic evaluation of how compositional effects influence convection and the conditions under which melting conditions may be reached. We model Enceladus’s south polar ice shell as a two-dimensional viscous layer in a spherical shell sector geometry. The computational domain is defined as a 2D annular sector with an angular opening of 180\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^\circ$$\end{document}, extending from the ice–ocean interface to the surface. Two ice shell thicknesses, 25 km and 50 km, are considered in order to evaluate the sensitivity of the thermal evolution to plausible end-member ice shell configurations. Although the present-day south polar ice shell is inferred to be significantly thinner (∼5–7 km), estimates based on gravity and libration data suggest a mean global shell thickness of ∼19–24 km, with strong lateral variations across Enceladus63. The selected thicknesses therefore span both present-day and plausible earlier or global-average configurations. The ice shell is treated as an incompressible fluid under the Boussinesq approximation, such that density variations are only included in the buoyancy term. Under tidal forcing conditions, deformation within the ice shell occurs at relatively low differential stresses, where ice behaves as a temperature–dependent viscous material. Although the rheology of ice near the melting point may involve multiple creep mechanisms (dislocation creep and grain boundary sliding), these processes are here represented through an effective viscosity formulation. This approach captures the dominant thermally controlled deformation behavior without explicitly resolving individual microscopic creep mechanisms. The numerical simulations are performed using the open-source geodynamic code ASPECT64–66, which solves the conservation equations of momentum, mass, and energy for thermally driven viscous flow,1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&\rho \frac{D \vec {v}}{D t} = - \nabla p + \eta \nabla ^2 \vec {v} + \rho \vec {g},\end{aligned}$$\end{document}2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&\frac{\partial T}{\partial t} + \vec {v} \cdot \nabla T = \kappa \nabla ^2 T + \frac{Q}{c},\end{aligned}$$\end{document}3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&\nabla \cdot \vec {v} = 0. \end{aligned}$$\end{document}where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\vec {v}$$\end{document} is the velocity, p is the pressure, T is the temperature, ρ is the density, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\rho _0$$\end{document} is the reference density, α is the coefficient of thermal expansion, η is the viscosity, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\vec {g}$$\end{document} is the gravitational acceleration acceleration, κ is the coefficient of thermal diffusion, c is the specific heat, and Q is the efficiency of heat sources. The mesh was initialized with a global refinement level of 4 and no adaptive refinement during runtime, providing uniform resolution sufficient to capture thermal boundary layers and convective cells throughout the shell. Velocity and temperature fields were discretized using fourth-order polynomial elements to balance accuracy and computational efficiency. Free-slip boundary conditions are implemented by imposing zero normal flow and allowing unrestricted tangential motion along all boundaries, consistent with expected behavior in planetary ice shells. Solver settings were configured to ensure robust convergence: the temperature field was solved with a relative tolerance of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{-10}$$\end{document} and the Stokes system was allowed up to 2000 iterations with a linear tolerance of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{-5}$$\end{document} per timestep, using a full block preconditioner to handle the coupled incompressible flow efficiently. It was computed a radial constant gravity field of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$0.113~\text {m/s}^2$$\end{document}, and a computational domain with a radius of 250 km. Convective behavior is characterized using the Rayleigh number (Ra), defined as the ratio of thermal buoyancy forces to viscous and diffusive effects,4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} \text {Ra} = \frac{\alpha \rho g \Delta T D^3}{\eta \kappa }, \end{aligned}$$\end{document}where α is the thermal expansivity, ρ is the density, g is the gravity acceleration, D is the thickness of the domain, Δ T is the super adiabatic temperature difference between the bottom and the top of the fluid, η is the viscosity of the well mixed interior, and κ is the thermal diffusivity. The surface temperature is fixed at 73 K, while the basal boundary is defined by a temperature-dependent melting condition derived from the ammonia–water ice phase relations. Mechanical boundary conditions are applied at both the top and bottom boundaries. The initial temperature field is defined as a conductive profile with superimposed perturbations to trigger convective instabilities. The viscosity is assumed to be temperature-dependent and follows an Arrhenius-type law,5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} \eta (T) = \eta _0 \exp \left[ \frac{E}{R}\left( \frac{1}{T} - \frac{1}{T_0}\right) \right], \end{aligned}$$\end{document}where η (T) is the dynamic viscosity at temperature T, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta _0$$\end{document} is the reference viscosity at the reference temperature \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$T_0$$\end{document}, E is the activation energy, and R is the universal gas constant. This formulation is consistent with previous geodynamic studies of ice-shell convection and provides a physically grounded description of temperature-dependent rheology. The reference viscosity \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta _0$$\end{document} is varied between \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{13}$$\end{document} Pa s, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{14}$$\end{document} Pa s, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{15}$$\end{document} Pa s to explore different rheological regimes62,67. The outer ice shell is expected to experience significant tidal heating due to periodic deformation induced by the satellite’s orbital motion. The spatial distribution and magnitude of tidal dissipation depend on the internal structure, orbital parameters, and rheological properties of the ice shell68,69. In particular, tidal heating is highly sensitive to the local viscosity, which in turn depends strongly on temperature. Within a viscoelastic framework, tidal dissipation arises from the phase lag between stress and strain during cyclic deformation. For ice near the melting point, the Maxwell relaxation time is on the order of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{4}$$\end{document} s – \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{5}$$\end{document} s, comparable to orbital forcing periods, implying that dissipation is maximized when viscous and elastic responses are of similar magnitude67,70. Under these conditions, the transition between elastic and viscous regimes controls the efficiency of energy dissipation71. Our system assumes tidal dissipation as the only internal heat source, implemented as a volumetric heating term derived from an incompressible Maxwell rheology. The dissipation rate depends primarily on the long-term viscosity described in Eq. (5) and the elastic shear modulus μ, and is expressed as,6\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} q = \frac{\epsilon _0^2 \, \omega ^2 \, \eta }{\left[ 1 + \left( \frac{\omega \eta }{\mu }\right) ^2 \right] }, \end{aligned}$$\end{document}where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\omega = 5.308 \times 10^{-5}~\text {s}^{-1}$$\end{document} is the tidal forcing frequency of Enceladus72, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mu = 3.3 \times 10^{9}~\text {Pa}$$\end{document} is the shear modulus of ice, and η is the dynamic viscosity. The parameter \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\epsilon _0$$\end{document} represents the tidal strain amplitude experienced by the ice shell. The tidal strain is approximated as \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\epsilon _0 = \xi / R$$\end{document}, where ξ is the vertical surface displacement amplitude during a tidal cycle and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$R = 250~\text {km}$$\end{document} is the radius of Enceladus. Using \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\xi \approx 0.8~\text {m}$$\end{document}72, we obtain \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\epsilon _0 \sim 3.2 \times 10^{-6}$$\end{document}. This formulation captures the transition between elastic and viscous regimes as a function of viscosity and tidal frequency. Thermodynamic models setup Pressure-dependent melting relations are used to describe the thermodynamic behavior of ammonia–water hydrates within the ice shell. The melting temperature of each composition is prescribed as a function of pressure, allowing the local melting threshold to increase with depth. The thermodynamic properties are parameterized following the semi–empirical framework of Ref.41, which combines laboratory measurements and interpolations between pure ice Ih and solid ammonia to estimate the behavior of mixed hydrates under pressures relevant to icy satellites. We note that this formulation represents a simplified description of ammonia–water ice thermodynamics. More comprehensive equations of state73 could provide a more detailed representation of phase behavior. However, the present approach captures the first–order thermodynamic effects relevant to this study, while more advanced formulations may refine quantitative predictions and will be explored in future work. Ammonia is assumed to originate from the subsurface ocean, where it acts as an antifreeze component. Convective transport and progressive freezing at the ice–ocean interface lead to the incorporation of ammonia hydrate phases into the overlying ice shell, allowing their thermodynamic influence to be evaluated within the modeled domain. In this framework, the ammonia–water ice system is represented using the ammonia monohydrate \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {NH}_{3}\cdot \hbox {H}_{2}\hbox {O}$$\end{document} and ammonia dihydrate \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {NH}_{3}\cdot \hbox {2H}_{2}\hbox {O}$$\end{document} phases. These two phases are specifically selected because they correspond to the dominant stable solid phases of the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {NH}_{3}\cdot \hbox {H}_{2}\hbox {O}$$\end{document} system under pressure–temperature conditions relevant to Enceladus’s ice shell, as constrained by experimental studies41. The volumetric thermal expansion of each solid phase is described by a power-law relation,7\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} \alpha (T) = A T^B \end{aligned}$$\end{document}where T is the absolute temperature and (A, B) are empirical constants derived from the parameterization. Integrating this relation yields an exponential formulation for the temperature-dependent density,8\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} \rho (T) = \rho _0 \exp \! \Big [ -\frac{A}{B+1} T^{B+1} \Big ] \end{aligned}$$\end{document}where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\rho _0$$\end{document} is the reference density for each compound. Following Ref.41, we adopt \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\rho _{0,\mathrm {NH_3\cdot 2H_2O}} = 982.6~\text {kg m}^{-3}$$\end{document} for ammonia dihydrate and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\rho _{0,\mathrm {NH_3\cdot H_2O}} = 958.8~\text {kg m}^{-3}$$\end{document} for ammonia monohydrate. The exponent B for monohydrate is taken from the parameterization proposed in Ref.41, where it is defined as the geometric mean between the thermal expansion exponents of ice Ih and pure solid NH\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$_3$$\end{document}. The coefficient A is constrained such that the resulting density–temperature relation is consistent with available experimental data, including X-ray diffraction measurements near 113 K. For both hydrates, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$A = 2.239 \times 10^{-7}$$\end{document} is adopted, ensuring consistency with the semi-empirical formulation and preserving thermodynamic coherence across the modeled temperature range. The melting temperature of each ammonia–water mixture is parameterized using a linear Clausius–Clapeyron relation based on experimentally derived slopes reported in Ref.41, Table III. For ammonia dihydrate (NH\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$_3\cdot$$\end{document}2H\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$_2$$\end{document}O), the peritectic melting point at near-atmospheric pressure is \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$T_m = 176.16$$\end{document} K, with a positive Clapeyron slope of dT/dP ≈ 0.259 K MPa\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{-1}$$\end{document}. This yields,9\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} T_{m,\mathrm {NH_3\cdot 2H_2O}} = 176.16 + 0.259\,P \end{aligned}$$\end{document}For ammonia monohydrate (NH\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$_3\cdot$$\end{document}H\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$_2$$\end{document}O), the congruent melting point near 1 bar is \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$T_m \approx 194.15$$\end{document} K, with a Clapeyron slope of dT/dP ≈ 0.460 K MPa\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{-1}$$\end{document}, yielding,10\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} T_{m,\mathrm {NH_3\cdot H_2O}} = 194.15 + 0.460\,P \end{aligned}$$\end{document}where P is the local hydrostatic pressure expressed in megapascals. These slopes arise directly from the thermodynamic Clausius–Clapeyron relation,11\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} \frac{dT}{dP} = \frac{\Delta S}{\Delta V} = \frac{L/T}{V_l - V_s}, \end{aligned}$$\end{document}where L is the latent heat of fusion, T is the absolute melting temperature, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V_l$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V_s$$\end{document} are the specific volumes of the liquid and solid phases, described in Ref.41, respectively. Although the Clapeyron slopes are derived from near-atmospheric pressure conditions, their application to pressures up to several megapascals, representative of Enceladus’ ice shell41, is could be justified by the absence of additional phase transitions within this pressure range57,58. As a result, the linear approximation is expected to remain valid to first order. Incorporating this pressure dependence ensures that melting temperatures increase consistently with depth, reflecting the stabilizing effect of compression on the hydrate crystal lattice. This avoids thermodynamic inconsistencies that would arise from neglecting pressure effects and provides a physically consistent framework for modeling the thermal evolution of ammonia-bearing ice shells. Melt generation proxy Melting is not explicitly resolved in the governing equations. Instead, a postprocessing approach is adopted to estimate the thermodynamic conditions under which partial melting conditions are reached. In this framework, the temperature field obtained from the geodynamic simulations is analyzed a posteriori, and thermal conditions exceeding the local pressure–dependent melting temperature are evaluated using a latent heat–based energetic proxy (Δ X). This proxy represents the excess thermal energy available above the solidus and provides a first–order thermodynamic measure of the proximity of the system to melting conditions, consistent with established formulations68. While this method does not resolve melt transport, segregation, or feedbacks between melt and rheology, it offers a physically meaningful approximation of where and when melting conditions are reached within the ice shell, following,12\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} \Delta X_{\text {melt}} = \frac{C_p}{L} (T - T_m) \Delta t^{-1}, \end{aligned}$$\end{document}where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$C_p$$\end{document} is the specific heat capacity, L the latent heat of fusion, and Δ t the timestep. Although more detailed phase diagrams would improve these estimates, the adopted approach could provide a physically consistent framework for modeling the thermal and compositional evolution of Enceladus’s outer ice shell under pressure conditions relevant to the system. The melt fraction (ϕ) was computed by integrating the volume of partially molten elements across the shell and normalizing by the total modeled shell volume (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V_{\text {shell}}$$\end{document}), as given by,13\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} \phi (t) = \frac{1}{V_{\text {shell}}} \sum _{i=1}^{N} \Delta X_i(t), \end{aligned}$$\end{document}where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta X_i(t)$$\end{document} represents the melt fraction within element i at time t, where each element contributes to the total melt fraction according to its local melt fraction and volume. This approach captures the spatiotemporal distribution of thermodynamic melting conditions driven by tidal dissipation and convective heat transport within Enceladus’s ice shell. The parameters used in the model are summarized in Table 1.Table 1General model parameters. SymbolMeaningValueUnit\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\rho _{0,\mathrm {NH_3\cdot 2H_2O}}$$\end{document}Reference density for ammonia dihydrate982.6kg m−3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\rho _{0,\mathrm {NH_3\cdot H_2O}}$$\end{document}Reference density for ammonia monohydrate958.8kg m−3αThermal expansion coefficient\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$2.239 \times 10^{-7}$$\end{document}K−1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$T_s$$\end{document}Surface temperature73K\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$T_{m,\mathrm {NH_3\cdot 2H_2O}}$$\end{document}Melting temperature for ammonia dihydrateEq. (9)K\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$T_{m,\mathrm {NH_3\cdot H_2O}}$$\end{document}Melting temperature for ammonia monohydrateEq. (10)KkThermal diffusivity\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$1.47 \times 10^{-6}$$\end{document}m2 s−1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$k_1$$\end{document}Thermal conductivity2.2W m−1 K−1ηViscosityEq. (5)Pa s\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta _0$$\end{document}Reference viscosity\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{13}$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{14}$$\end{document}, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{15}$$\end{document}Pa sRGas constant8.314J mol−1 K−1DIce shell thickness25 and 50kmgGravitational acceleration0.113m s−2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$L_{\mathrm {NH_3\cdot 2H_2O}}$$\end{document}Latent heat for ammonia dihydrate131.9kJ kg−1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$L_{\mathrm {NH_3\cdot H_2O}}$$\end{document}Latent heat for ammonia monohydrate187.2kJ kg−1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$C_p$$\end{document}Specific heat capacity2000J kg−1 K−1AThermal expansion coefficient for both hydrates\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\approx 2.239 \times 10^{-7}$$\end{document}−BMean thermal expansion exponent for both hydrates≈ 1.375−Data sources are cited in the main text. Table 1General model parameters. SymbolMeaningValueUnit\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\rho _{0,\mathrm {NH_3\cdot 2H_2O}}$$\end{document}Reference density for ammonia dihydrate982.6kg m−3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\rho _{0,\mathrm {NH_3\cdot H_2O}}$$\end{document}Reference density for ammonia monohydrate958.8kg m−3αThermal expansion coefficient\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$2.239 \times 10^{-7}$$\end{document}K−1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$T_s$$\end{document}Surface temperature73K\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$T_{m,\mathrm {NH_3\cdot 2H_2O}}$$\end{document}Melting temperature for ammonia dihydrateEq. (9)K\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$T_{m,\mathrm {NH_3\cdot H_2O}}$$\end{document}Melting temperature for ammonia monohydrateEq. (10)KkThermal diffusivity\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$1.47 \times 10^{-6}$$\end{document}m2 s−1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$k_1$$\end{document}Thermal conductivity2.2W m−1 K−1ηViscosityEq. (5)Pa s\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta _0$$\end{document}Reference viscosity\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{13}$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{14}$$\end{document}, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{15}$$\end{document}Pa sRGas constant8.314J mol−1 K−1DIce shell thickness25 and 50kmgGravitational acceleration0.113m s−2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$L_{\mathrm {NH_3\cdot 2H_2O}}$$\end{document}Latent heat for ammonia dihydrate131.9kJ kg−1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$L_{\mathrm {NH_3\cdot H_2O}}$$\end{document}Latent heat for ammonia monohydrate187.2kJ kg−1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$C_p$$\end{document}Specific heat capacity2000J kg−1 K−1AThermal expansion coefficient for both hydrates\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\approx 2.239 \times 10^{-7}$$\end{document}−BMean thermal expansion exponent for both hydrates≈ 1.375−Data sources are cited in the main text. General model parameters. Data sources are cited in the main text. Supplementary Information Supplementary Information. Supplementary Information.