Ice melts at 0 °C (32 °F) under normal atmospheric pressure, a value so fundamental it was once used to define the Celsius scale itself. That single number, however, only tells you what happens to a chunk of pure ice sitting on your kitchen counter. Change the pressure, dissolve something in the surrounding water, shrink the ice down to nanometer-scale pores, or apply a strong electric field, and the melting point shifts, sometimes dramatically. Even the surface of an ordinary ice cube is not quite solid at temperatures well below freezing.
Why the Familiar Number Is 0 °C
At standard atmospheric pressure (about 101 kPa), the molecules in an ice crystal vibrate in a rigid hexagonal lattice held together by hydrogen bonds. When you add enough thermal energy to disrupt that lattice faster than it can re-form, the solid turns to liquid. The temperature at which the rate of melting and the rate of freezing reach equilibrium is, for pure water at sea-level pressure, 0 °C. This is a thermodynamic balance point: below it, freezing wins; above it, melting wins. The number is not arbitrary, but it only applies when the water is pure, the pressure is roughly one atmosphere, and the ice exists in bulk rather than in microscopic crevices or thin films.
How Salt and Other Solutes Push the Melting Point Down
Tossing salt on an icy sidewalk works because dissolved ions interfere with the orderly process of water molecules locking back into the crystal lattice. When sodium chloride dissolves in the thin film of liquid water on ice, it lowers the chemical potential of the liquid phase, meaning the ice has to get colder before it can maintain its solid structure. Molecular dynamics simulations of NaCl on ice confirm that the freezing-point depression observed matches experimental measurements well.1PubMed. The effect of salt on the melting of ice: A molecular dynamics simulation study The magnitude of the depression depends on the concentration of dissolved solute. For common road salt, the practical limit is somewhere around −18 °C (0 °F); below that temperature, the salt can no longer keep enough liquid water around to dissolve itself, and it stops being effective.
Salt is not unique in this regard. Any dissolved substance, from sugar to calcium chloride to antifreeze, depresses the melting point by roughly the same mechanism. Calcium chloride releases three ions per formula unit compared to sodium chloride’s two, which is why it works at lower temperatures and is preferred for extremely cold climates. The underlying principle is the same: more dissolved particles in the liquid phase means the ice has to be colder to survive.
The Surface That Is Already Liquid Below Freezing
One of the stranger facts about ice is that its outermost surface is not fully solid, even at temperatures below 0 °C. A thin layer of disordered, liquid-like water exists on the surface of ice crystals at sub-freezing temperatures. Scientists call these quasi-liquid layers, and they govern many familiar phenomena: the slipperiness of ice, the way snowflakes stick together, and the sintering of snow into firn in glaciers.2PubMed Central. Two types of quasi-liquid layers on ice crystals are formed kinetically
Advanced optical microscopy has shown that these layers come in at least two distinct phases with different shapes and dynamics. Researchers directly visualized these phases on ice crystal surfaces, resolving individual molecular steps just 0.37 nanometers thick.3PubMed Central. Quasi-liquid layers on ice crystal surfaces are made up of two different phases The layers grow thicker as the temperature approaches 0 °C, and they are thicker where the crystal surface is less orderly. This means that the boundary between “ice” and “liquid water” is not a clean line but a gradient, especially near the melting point. If you have ever noticed that ice feels slippery even on a day well below freezing, the quasi-liquid layer is a big part of the reason.
What Pressure Does to the Melting Point
Water is one of the rare substances where the solid is less dense than the liquid. Because ice floats, squeezing it tends to favor the denser liquid state, which means that moderate increases in pressure actually lower the melting point slightly. Under everyday conditions the effect is small: the pressure under an ice skate blade, for instance, shifts the melting point by a fraction of a degree at most, far less than old textbooks used to claim. The “pressure melting” explanation for ice skating is largely a myth; the quasi-liquid surface layer and friction are far more important.
At extreme pressures found deep inside planets, the story changes completely. Ice takes on exotic crystal structures very different from the hexagonal ice we know. First-principles simulations of ice-VII, a high-pressure phase, have mapped its melting curve between 10 and 50 gigapascals, finding results consistent with diamond-anvil-cell experiments.4PubMed Central. Melting of ice under pressure At pressures above roughly 45 GPa, the melting curve steepens sharply because the ice molecules begin to dissociate and protons start diffusing through the solid before it melts, behavior resembling what physicists call a superionic state. At those pressures, the “melting temperature” of ice can climb to hundreds or even thousands of degrees Celsius, a world away from 0 °C.
Tiny Spaces Change the Rules
If you confine water inside pores just a few nanometers wide, the melting point drops substantially. This is not because of impurities or pressure, but because the curvature of the solid-liquid interface in a tiny pore adds an energy penalty that destabilizes the crystal. The smaller the pore, the larger the depression. Molecular simulations confirm that the melting temperature of nanoconfined ice is strongly depressed relative to bulk ice, depends only weakly on how full the pore is, and is surprisingly insensitive to whether the pore wall is water-attracting or water-repelling.5PubMed. Melting and crystallization of ice in partially filled nanopores
Experiments using porous silica materials with well-defined pore sizes show this relationship follows a simple inverse rule: the melting-point depression scales linearly with the inverse of the effective pore radius. A roughly 0.4-nanometer-thick layer of water molecules bonds to the pore wall and never crystallizes, effectively shrinking the available space for ice.6Scientific Reports. Layer-by-Layer Freezing of Nanoconfined Water In pores just two or three nanometers across, the melting point can drop by tens of degrees. This matters in geological and biological contexts: water in the nanopores of rocks, soils, and biological membranes can remain liquid at temperatures where bulk water would be frozen solid. It also matters in industrial applications like catalysis and desalination membranes, where understanding phase behavior in tiny channels is essential.
Grain Boundaries in Polycrystalline Ice
Real-world ice, whether in a glacier or a freezer tray, is rarely a perfect single crystal. It is made of many small grains with different crystallographic orientations, and the boundaries where those grains meet behave differently from the bulk. Even in pure ice, grain boundaries have a measurable liquid-like thickness, and that thickness depends on the misorientation angle between the two grains. When salt is present in the surrounding liquid, grain-boundary thickness increases further, from roughly 1 nanometer in pure ice to about 8 nanometers at moderate salt concentrations.7arXiv. Grain boundary melting in ice
This grain-boundary premelting has consequences for everything from the mechanical strength of sea ice to the transport of brine through glacial ice. Sea ice, which forms from salt water, traps pockets and channels of brine within a matrix of relatively pure ice crystals. As the ice warms toward its melting point, these grain-boundary channels widen, making the ice more permeable and more fragile. The internal “plumbing” of sea ice is governed not just by bulk melting at 0 °C but by the nanoscale premelting that happens along every crystal boundary at lower temperatures.
Can You Superheat Ice Past 0 °C Without Melting It?
In principle, melting should begin the instant ice crosses 0 °C, but in practice the transition is not instantaneous. Researchers have managed to superheat ice, pushing it above 0 °C for brief periods without it fully converting to liquid. In one experiment, ice at an initial temperature of −3 °C was hit with an ultrafast infrared laser pulse that excited the water molecules’ stretching vibrations. The result was an average temperature jump of about 20 K that persisted for at least 250 picoseconds without melting.8PubMed. Ultrafast superheating and melting of bulk ice
Follow-up work tracked the full dynamics of superheating and melting over timescales from tens of picoseconds to microseconds. Even when a large excess of energy was dumped in, only about 30 percent of a micrometer-scale ice crystal melted within the first 20 to 25 nanoseconds, because the superheated solid persisted as a metastable phase for more than 100 nanoseconds. The energy was trapped in low-energy vibrational states that favored a disordered but still solid-like “plastic” ice phase rather than immediate liquid water.9PubMed. Superheating and Homogeneous Melting Dynamics of Bulk Ice You will never notice this delay while watching an ice cube melt on the counter, because nanoseconds are far too short for human perception. But for physicists studying phase transitions, the existence of superheated ice reveals that melting is not a simple on-off switch but a kinetic process with its own bottlenecks and barriers.
Electric Fields and the Melting of Ice
Electric fields add another unexpected twist. The conventional wisdom used to be that electric fields help water freeze by aligning molecules into orderly arrangements, and for some ice structures that is true. But molecular dynamics simulations have shown that for a monolayer of ice confined between two flat plates, applying an electric field perpendicular to the layer actually melts it. The field disrupts the hydrogen-bond network that the confining walls had helped stabilize, lowering the melting temperature as field strength increases.10PubMed. Electromelting of confined monolayer ice
For bulk ice, the picture is more nuanced. Simulations of ice Ih, the ordinary hexagonal ice, show that its melting temperature, crystal growth speed, and nucleation rate all decrease as an applied electric field strengthens. But a different crystal form, polarized cubic ice (ice Icf), is actually stabilized by a strong field, and its melting point rises with field intensity.11Journal of Physics: Condensed Matter. Phase boundaries, nucleation rates and speed of crystal growth of the water-to-ice transition under an electric field: a simulation study In other words, the effect of an electric field on ice depends entirely on the crystal structure and the geometry of confinement. There is no single answer to “does electricity melt ice or freeze water” because different ice phases respond in opposite ways.
How Living Organisms Manipulate Ice’s Melting Point
Some organisms have evolved proteins that directly interfere with ice crystal growth, effectively decoupling the melting point from the freezing point. Antifreeze proteins, found in certain fish, insects, plants, and bacteria, bind to the surface of tiny ice crystals and block new water molecules from joining the lattice. This creates a gap between the temperature at which ice melts and the temperature at which it grows, a phenomenon called thermal hysteresis. The melting point stays essentially unchanged, but the freezing point drops by anywhere from a fraction of a degree (in most fish) to several degrees (in some insects), depending on the protein’s size and the area of its ice-binding surface.12Elsevier. Thermal hysteresis activity of antifreeze proteins: A model based on fractional statistics theory of adsorption
This is a clever trick: rather than changing the thermodynamic melting point of ice (which would require altering the water itself), the proteins exploit kinetics. They prevent ice from growing without preventing it from melting. The result is that an Arctic fish can swim in water at −1.8 °C, well below the normal freezing point of its body fluids, because even though ice crystals form, they cannot grow large enough to cause damage. Researchers have explored synthetic versions of these proteins for applications ranging from preserving transplant organs to improving the texture of frozen food.
Amorphous Ice and Its Glass Transition
Not all solid water is crystalline. If you cool water quickly enough, or deposit it from vapor at very low temperatures, you can create amorphous ice, a glassy solid with no regular crystal lattice. Amorphous ice does not have a conventional melting point in the usual sense. Instead, it undergoes a glass transition: a gradual shift from a brittle, rigid glass to a more fluid state. For high-density amorphous ice, careful calorimetry experiments have identified a repeatable, reversible glass transition at about 116 K (−157 °C).13Journal of Non-Crystalline Solids. The glass transition in high-density amorphous ice
This is far below the temperature where crystalline ice melts, and it represents a fundamentally different kind of transition. In a glass transition, the material softens and its molecules gain enough mobility to rearrange, but there is no latent heat release and no sharp dividing line between solid and liquid. Amorphous ice matters in astrophysics, where water ice in comets and on the surfaces of moons in the outer solar system likely exists in amorphous or partially crystalline forms at temperatures well below 116 K. When those ices warm, their transformation behavior is governed by glass transitions and crystallization rather than simple melting.
Real-World Melting Is Messier Than a Thermometer Suggests
In nature, ice rarely melts at a clean threshold temperature. An eleven-year study of surface energy balance at Taylor Glacier in Antarctica found that air temperatures above freezing did not always produce melt, and melt sometimes occurred when air temperatures were below freezing.14Journal of Geophysical Research: Earth Surface. Surface energy balance and melt thresholds over 11 years at Taylor Glacier, Antarctica The reason is that melting depends on the total energy balance at the ice surface, not just air temperature. Solar radiation, wind speed, humidity, and the ice’s own albedo all contribute. On a calm, sunny day with air at −2 °C, strong solar absorption can deliver enough energy to the surface to drive melting. On a windy, cloudy day at +1 °C, convective cooling and low radiation might prevent it.
This mismatch between air temperature and actual melting is critical for climate science. Models that estimate glacier and ice-sheet melt purely from air temperature miss important feedbacks. Darkened ice surfaces from soot or algae growth absorb more sunlight and melt faster than clean ice at the same temperature. Wind-driven sublimation can cool the surface and suppress melting. The 0 °C number remains the thermodynamic anchor, but the real question for any particular piece of ice is whether the net energy flux at its surface is positive enough to overcome the latent heat of fusion, and that depends on a whole suite of environmental conditions beyond what a thermometer reads.
When “Melting Point” Means Something Different on Other Worlds
On Earth, we think of ice as a single familiar substance, but water ice in the solar system exists under conditions so extreme that the concept of a melting point gets stretched. On the icy moons of Jupiter and Saturn, pressures in their interiors can reach tens of gigapascals, stabilizing high-pressure ice phases whose melting temperatures are hundreds of degrees above 0 °C. As noted earlier, ice-VII at pressures above 45 GPa begins to exhibit proton diffusion before melting, a superionic state where the oxygen atoms hold a crystal lattice while hydrogen ions flow freely through it.4PubMed Central. Melting of ice under pressure This means that deep inside a moon like Ganymede, there could be layers of hot, electrically conductive ice that are simultaneously solid (in terms of the oxygen framework) and liquid (in terms of proton transport).
Meanwhile, on comet surfaces and in the cold reaches of the outer solar system, temperatures hover at 30 to 80 K, and ice exists in amorphous or exotic crystalline forms that never encounter their melting point under natural conditions. When a comet swings close to the Sun, its surface ice sublimates directly into vapor rather than melting, because the ambient pressure is essentially zero. The concept of a melting temperature requires both the right temperature and the right pressure; in vacuum, the liquid phase of water barely exists at any temperature. For these ices, the relevant transition is sublimation, and the energy needed to drive it comes almost entirely from solar radiation, not from ambient warmth.