🧊
Some ice is soft and some ice is strong. The ice in your cup is one kind. Scientists found more than twenty kinds. One kind, ice four, is shy. It hides and changes fast. Ice is full of surprises.
What Is Ice? 🧊
Ice is frozen water. The ice in your drink is called ice one. It is the only ice most people ever see.
Can Ice Be Hot? 🔥
Yes. Scientists squeeze water very hard and it turns into ice, even when it is warm. This ice is called ice seven. It is much stronger than freezer ice.
What Is Ghost Ice? 👻
Ice four is the shy one. It appears for a tiny moment, then changes into a different ice. Scientists call this metastable. That is a big word for something that cannot sit still.
One Ice Cube, Twenty Ices 🧊
The ice in your drink is called ice Ih. Say it like "ice one h." It is the ice in freezers, snowmen, and every snowflake you have caught. For a long time, people thought it was the only ice. They were wrong. Scientists have found more than twenty kinds of ice, each with its own hidden shape, including a cubic cousin of Ih called ice Ic.
What Makes the Ices Different?
Zoom in close and ice is a crowd of water molecules holding hands. How they link up decides which ice you get. In ice Ih, the molecules form open six-sided rings, like a honeycomb. That open shape is why ice floats: it takes up more room than liquid water.
Can Ice Be Hot? 🔥
Yes. Squeeze water hard enough and it freezes even at room temperature. Ice VII forms this way, and scientists call it "hot ice." In 2018, researchers found tiny pieces of ice VII trapped inside real diamonds, carried up from hundreds of kilometers deep in the Earth, where the pressure is enormous.
What Is the Ghost Ice? 👻
Ice IV is the shy one. Every other ice has its own patch of temperature and pressure where it is most stable. Ice IV has no patch of its own. It appears briefly inside the territory of ice V, then changes into something else. Scientists call this metastable, a big word for "not built to last." One writer called it the will-o'-the-wisp, after ghostly lights from old stories.
Where Else Do Strange Ices Live? 🪐
Far from Earth. The moons Ganymede and Titan hide deep oceans under their frozen surfaces, and at the bottom the pressure is high enough to make ices II, III, V, and VI. Inside Uranus and Neptune sits ice XVIII, where oxygen atoms stay frozen while hydrogen flows through them like water. It is hotter than lava and still called ice.
Every ice cube you have ever touched is the same kind of ice. Scientists call it ice Ih, where the h stands for hexagonal; a rarer cubic cousin, ice Ic, exists too. Its molecules link into six-sided rings, which is why snowflakes grow six arms. But Ih is only the first entry in a longer catalog. Change the temperature, squeeze the water hard enough, and the molecules snap into entirely new arrangements; the open honeycomb that feels so comfortable at normal pressure cannot survive the squeezing, so the molecules collapse into tighter and denser packings with every added atmosphere; and each new packing is a different ice, with its own crystal shape and density. Researchers have found more than twenty phases, from ice I to ice XXI, including one ghost phase with no home: ice IV.
What Decides Which Kind of Ice You Get?
Two things: temperature and pressure. At normal pressure, cold water becomes familiar ice Ih. Squeeze water to about a thousand times normal air pressure while cooling it, and ice III forms instead, with its ordered cousin ice II nearby. Squeeze harder to march through ice V and ice VI, each packing tighter. The map of which phase wins where is called a phase diagram, and water's is among the most crowded in all of chemistry; more than twenty crystalline phases jostle for space alongside several glassy forms of ice that have no crystal structure at all, each one holding its own corner of temperature and pressure.
Who Found All These Ices?
In 1912, Percy Bridgman built machines that squeezed water harder than ever before. In one year he discovered four new ices: II, III, V, and VI. His high-pressure work earned the 1946 Nobel Prize. The catalog kept growing through the twentieth century and into the twenty-first, reaching ice XXI in 2025; a Korean team filmed freezing water with X-ray movies fast enough to catch a brand-new phase crystallizing in real time, proving that every generation of better instruments finds new ices hiding in water's crowded phase diagram.
Why Is Ice IV Called the Ghost Phase?
Every other ice owns a region of the phase diagram where it is the most stable arrangement, but ice IV owns nothing. In 1972, Engelhardt and Whalley mapped where it appears: 4,100 to 6,000 times normal air pressure, at minus 18 to minus 7 degrees Celsius, inside ice V's territory. It visits briefly, then transforms into a neighbor. Scientists call this metastable: stable enough to study for now. In 1981, Engelhardt and Barclay Kamb solved its structure anyway; they found a rhombohedral crystal with 16 water molecules per repeating unit and a feature never seen before in any ice, a hydrogen bond passing straight through the center of a six-sided ring of molecules, stitching together layers that do not even touch. One writer nicknamed it the will-o'-the-wisp of physics.
Can Ice Really Be Hot?
Yes, and you can make it in a lab. Squeeze water to 1.1 gigapascals at room temperature and it freezes into ice VI. Past about 2 gigapascals you get ice VII, called hot ice because it forms where water should be liquid. In 2018, scientists found ice VII crystals trapped inside diamonds carried up from hundreds of kilometers deep. Whole oceans of ices VI and VII may sit at the bottom of the seas on the moons Ganymede and Titan; there, water meets rock under crushing pressure, and studying those hidden oceans is one of the biggest reasons scientists care so much about mapping the high-pressure ices.
What Is the Strangest Ice of Them All?
Meet ice XVIII, discovered in 2019. It exists above 2,000 degrees Celsius and a million atmospheres, deep inside Uranus and Neptune. There, oxygen atoms lock into a crystal while hydrogen atoms flow through freely, carrying current like a metal: a superionic phase. It is technically ice, and it is hotter than lava.
Are There More Ices Still Left to Find?
Almost certainly: in 2025, a Korean team caught ice XXI crystallizing on X-ray movies, a metastable phase with 152 water molecules per repeating unit, the most complex ice ever measured; and in 2026, a Tokyo group reported ice XXII in a preprint that other scientists are still checking. Water keeps surprising the people who study it most.
Water has the most crowded phase diagram of any common substance. More than twenty crystalline phases have been identified, from hexagonal ice Ih and its cubic cousin ice Ic to the superionic ices. Each phase is a compromise between two urges: the molecules want to form open networks of hydrogen bonds, while pressure wants them packed as tightly as possible; and everything strange about ice, from cubes that float to crystals hotter than lava, comes from the tension between those two demands. This level tours the diagram as a physicist sees it. We will use two equations, meet the ghost phase, and see why the count keeps climbing.
Why Ice Ih Melts Under Pressure
For most substances, squeezing raises the melting point. Ice Ih does the opposite: pressure lowers its melting point. The reason is density. Liquid water is denser than ice Ih, so pressure favors liquid. The relationship is exact, and it is written in the Clausius-Clapeyron equation.
Here L is the latent heat of melting, T is the temperature, and ΔV is the volume change on melting. For ice Ih, ΔV is negative, so the melting curve slopes downward as pressure rises.
That downward slope has everyday consequences; a skate blade concentrates a skater's weight onto a thin edge, and the pressure can melt a microscopic film of water beneath it, which is part of why skates glide, while the same physics lets glaciers slide over bedrock. It also explains the classic demo: a weighted wire cuts slowly through an ice block, with meltwater refreezing above as pressure releases behind it.
Proton Order and Disorder
Every water molecule is an oxygen atom with two hydrogens in a bent arrangement, each bonded to four neighbors. The oxygen atoms define the crystal lattice, but the hydrogens can point in different directions while the oxygens stay fixed; when those orientations are random, the phase is called proton-disordered, and when they lock into a repeating pattern, it is proton-ordered, which changes the electrical properties without changing the crystal shape. Several ices come in disordered-ordered pairs: Ih with XI below 72 kelvin, III with IX, V with XIII, VII with VIII, XII with XIV, and VI with two ordered relatives, XV and XIX; ice II is an ordered phase with no disordered twin. The ordered forms appear only at very low temperatures, where molecular tumbling slows enough for the hydrogens to settle; but reorientation is so sluggish down there that experimenters dope the ice with traces of hydrochloric acid to speed it up, which means the newest ordered phases were all discovered in doped samples, and researchers still debate which features are truly intrinsic. The doping trick works, but keeps the field honest.
Deep Dive: Ice IV, the Phase With No Territory
In 1981, Engelhardt and Barclay Kamb solved the structure despite the difficulty; they found a rhombohedral crystal with 16 water molecules per unit cell and a density of 1.27 grams per cubic centimeter at 110 kelvin, plus a feature seen in no other ice, a hydrogen bond passing straight through the center of a six-membered ring to link layers that are not adjacent. Even quenched to low temperature, the phase stays proton-disordered.
Making it on demand took until 2001, when Salzmann and coworkers published a recipe: warm high-density amorphous ice at 0.81 gigapascals, heating slowly at 0.4 kelvin per minute, and ice IV crystallizes near 165 kelvin; heat faster than about 10 kelvin per minute and ice XII forms instead, which explains why the phase hid for so long. The narrow recipe also explains the Engelhardt-Kamb collapse hypothesis: squeezing ordinary ice might convert it directly into ice IV, an idea supported by ammonium fluoride experiments.
Ostwald's Step Rule and the Discovery of Ice XXI
In 2025, a team led by Lee Geun Woo at the Korea Research Institute of Standards and Science watched water crystallize with an X-ray microscope. They used a diamond anvil cell that squeezes in 10 milliseconds instead of tens of seconds, which let them supercompress liquid water to more than twice the pressure where it would normally freeze while keeping it liquid at room temperature; then they filmed the freezing with the European X-ray free-electron laser, capturing images every microsecond. The water refused the direct path: instead of jumping straight to ice VI, the stable phase, it passed through the new metastable ice XXI first. Ice XXI has a body-centered tetragonal unit cell holding 152 water molecules, and only after visiting it did the sample settle toward ice VII. This is Ostwald's step rule, named for the chemist Wilhelm Ostwald: a crystallizing system visits the most reachable phases first, not the most stable; ice IV is the classic illustration, crystallizing inside ice V's territory because its structure is the easiest arrangement for the molecules to reach. The 2025 experiment showed the same rule operating at room temperature, on camera.
Ices of the Outer Solar System
The phase diagram is written across the solar system; Ganymede, the largest moon in the solar system, hides a deep ocean under its ice shell, and the pressure at the ocean floor favors ices III, V, and VI, while Titan is similar. On Earth, high-pressure ice is not just a lab product: in 2018, Tschauner and colleagues found ice VII trapped as inclusions in natural diamonds from the mantle transition zone. Farther out, inside Uranus and Neptune, water is expected to exist as ice XVIII, the superionic phase first produced in 2019 by Millot and colleagues; there the oxygen atoms sit in a face-centered cubic lattice while hydrogen ions flow through it like a liquid, conducting electricity, which may explain the planets' odd magnetic fields.
Pressure at depth h in a fluid of density ρ. For water, each 10 meters of depth adds about 1 atmosphere, so ice VI's 11,000 atmospheres correspond to roughly 110 kilometers of ocean.
How the Count Keeps Climbing
New phases keep appearing because better instruments keep revealing faster and stranger crystallization paths; ice XVI, reported in 2014, was made by pumping gas out of a clathrate hydrate, leaving a crystal that is mostly empty cage, the least dense ice known at just 0.81 grams per cubic centimeter. Ice XVII, from 2016, came from a hydrogen-filled crystal that can soak hydrogen back up. In 2023, ball-milling ice at liquid-nitrogen temperature produced medium-density amorphous ice, a disordered form that may exist on interstellar dust grains. In 2026, a University of Tokyo group reported ice XXII in a preprint, with a repeating unit of 304 molecules. A preprint has not passed peer review, so ice XXII is a candidate, not a confirmed phase; still, the pattern is clear: every faster camera or gentler squeeze finds water a new way to be ice.
No substance has been studied more than water, and none keeps stranger secrets. Under the right temperature and pressure, water freezes into more than twenty distinct crystalline phases, plus a family of amorphous forms with no crystal structure at all. The ice in your freezer, ice Ih, is merely the phase that happens to be stable at one atmosphere and below freezing. Everywhere else in the diagram, water finds other ways to be solid, and the inventory is still growing.
A Century of Squeezing
In 1912, the Harvard physicist Percy Bridgman published the first systematic map of water under pressure and announced four new ices: II, III, V, and VI. Ice I was the familiar kind; ice IV would prove far more elusive than its number suggests. Bridgman's apparatus work earned him the 1946 Nobel Prize in Physics, and his numbering scheme stuck: new phases are still numbered with Roman numerals in order of discovery, with ice I split into hexagonal Ih and cubic Ic.
Even ice I comes in two forms: hexagonal Ih and cubic Ic, though for decades every laboratory 'cubic ice' was really a stacking-disordered mixture, and pure cubic ice was only synthesized in 2020 by warming porous ice XVII. Ice II, one of Bridgman's 1912 finds, is a fully ordered rhombohedral crystal. The dense cubic phases, ices VII and VIII, followed in the mid-twentieth century, with VIII the proton-ordered form of VII. Ice IX turned out to be real, though nothing like Kurt Vonnegut's fiction: it is stable only below about 140 kelvin and high pressure. Ice X, above roughly 60 gigapascals, is where hydrogen bonds become symmetric, the proton sitting exactly midway between two oxygens. Ice XI, the proton-ordered form of ordinary ice, appears below 72 kelvin. Then came the modern era: ice XII, solved in 1998 by Lobban, Finney, and Kuhs; the proton-ordered phases XIII and XIV in 2006 and XV in 2009, all from Salzmann and colleagues; ice XVI in 2014, made by emptying a gas hydrate to leave the least dense crystalline ice known; ice XVII in 2016, from a hydrogen-filled crystal; ice XVIII in 2019, the superionic phase, caught with nanosecond X-ray diffraction of shock-compressed water; ice XIX in 2021, shown to be a second proton-ordered relative of ice VI, distinct from ice XV; ice XX, a second superionic phase with a body-centered cubic oxygen lattice, mapped by Prakapenka and colleagues in diamond-anvil experiments; and in 2025, ice XXI, a metastable tetragonal phase with 152 molecules per unit cell, found by the KRISS team filming crystallization with microsecond X-ray movies.
Metastability: The Ghost in the Diagram
Most of these phases own a region of the phase diagram where they are the thermodynamically stable choice. A few do not. Ice IV, characterized by Engelhardt and Whalley in 1972, forms between about 4,100 and 6,000 bar at minus 18 to minus 7 degrees Celsius, squarely inside the stability field of ice V. It is metastable with respect to ices III, V, and VI: every neighbor is more stable, yet ice IV appears anyway, briefly, before converting.
When Engelhardt and Barclay Kamb solved its structure in 1981, they found a rhombohedral crystal with 16 molecules per unit cell and a density of 1.27 grams per cubic centimeter at 110 kelvin and ambient pressure, plus a structural oddity unique in ice science: a hydrogen bond threading straight through a six-membered ring to join non-adjacent layers. The phase remains proton-disordered even when quenched. A reproducible preparation arrived only in 2001, when slow warming of high-density amorphous ice at 0.81 gigapascals was shown to crystallize ice IV near 165 kelvin, while faster heating yields ice XII instead.
Ice IV matters because it is the clearest proof of Ostwald's step rule in water: crystallization favors the most accessible phase, not the most stable. Squeezed or supercooled water reaches ice IV's arrangement first, then relaxes toward true equilibrium. The same rule governed the 2025 observation of ice XXI as a transient between supercompressed liquid and ice VII. A science writer's nickname for ice IV, the will-o'-the-wisp, is unusually apt: it is real, it glows briefly in the data, and it is gone before you can be sure you saw it.
Order, Disorder, and the Doping Question
Several ices exist as disordered-ordered pairs, where the oxygen lattice is identical but the hydrogens go from random orientations to a locked pattern at low temperature. The pairs now include Ih-XI, V-XIII, XII-XIV, VI-XV, and VI-XIX. The catch is kinetics: molecular reorientation slows to a crawl at the temperatures where ordering sets in, so experimenters add traces of hydrochloric acid to catalyze it. The trick works, but it means the newest ordered phases were all discovered in doped samples, and the community continues to scrutinize which features are intrinsic. It is a healthy argument, and it is not settled.
Planetary Stakes
The diagram is written across the solar system. Ganymede and Titan likely sandwich their subsurface oceans between ice Ih above and high-pressure ices below, and the properties of those deep ices control whether the oceans can exchange chemistry with the rocky seafloor. On Earth, high-pressure ice occurs naturally: Tschauner and colleagues found ice VII as inclusions in diamonds from the mantle transition zone, the first natural sample. Inside Uranus and Neptune, water is expected to be superionic ice XVIII, with mobile protons conducting electricity through a frozen oxygen lattice, a state that may shape the planets' odd magnetic fields.
What Remains Unknown
In 2026, a University of Tokyo group reported ice XXII in a preprint, with a 304-molecule repeating unit. Until peer review, it is a candidate. The amorphous side is equally lively: the 2023 discovery of medium-density amorphous ice, made by ball-milling at 77 kelvin, suggests the glassy states of water are as varied as the crystals. And the deepest question, whether liquid water itself splits into two distinct liquids at low temperature, remains open. A century after Bridgman, the phase diagram of the most familiar substance on Earth is still being drawn.
Sources
- Bridgman, P. W. (1912). "Water, in the Liquid and Five Solid Forms." Proceedings of the American Academy of Arts and Sciences, 47, 441-558.
- Engelhardt, H. & Whalley, E. (1972). "Ice IV." The Journal of Chemical Physics, 57, 2678-2684.
- Engelhardt, H. & Kamb, B. (1981). "Structure of ice IV, a metastable high-pressure phase." The Journal of Chemical Physics, 75, 5887-5899.
- Lobban, C., Finney, J. L. & Kuhs, W. F. (1998). "The structure of a new phase of ice." Nature, 391, 268-270.
- Salzmann, C. G. et al. (2006). "The preparation and structures of hydrogen ordered phases of ice." Science, 311, 1758-1761.
- Falenty, A., Hansen, T. C. & Kuhs, W. F. (2014). "Formation and properties of ice XVI obtained by emptying a type sII clathrate hydrate." Nature, 516, 231-233.
- Tschauner, O. et al. (2018). "Ice-VII inclusions in diamonds: Evidence for aqueous fluid in Earth's deep mantle." Science, 359, 1136-1139.
- Millot, M. et al. (2019). "Nanosecond X-ray diffraction of shock-compressed superionic water ice." Nature, 569, 251-255.
- Gasser, T. M. et al. (2021). "Structural characterization of ice XIX as the second polymorph related to ice VI." Nature Communications, 12, 1128.
- Rosu-Finsen, A., Davies, M. B. & Salzmann, C. G. (2023). "Medium-density amorphous ice." Science, 379, 474-478.
- Algara-Siller, G. et al. (2015). "Square ice in graphene nanocapillaries." Nature, 519, 443-445.
- Vonnegut, K. (1963). Cat's Cradle. New York: Holt, Rinehart and Winston.
- Lee, G. W. and colleagues, Korea Research Institute of Standards and Science (2025). Discovery of metastable ice XXI by dynamic diamond-anvil-cell compression, reported in Nature Materials. Body-centered tetragonal, 152 water molecules per unit cell.
- Quanta Magazine (2026). "Physicists Discover the Most Complex Forms of Ice Yet." Reports the ice XXII preprint (304-molecule repeat) and the "will-o'-the-wisp" nickname for ice IV.