Explore Phase Changes with AI-Assisted Molecular Simulations
From the silicon in your phone to the salt on your table: grow crystals from their melts and investigate with an AI agent.
Chemistry textbooks sort crystalline solids into four families — metallic, ionic, covalent, and molecular — and explain that the type of bond holding a crystal together decides how hard it is to take apart. Students are usually given a table of melting points to memorize. With AIMS, students investigate the connection between bonding and melting. They talk to an AI agent, and it runs real molecular dynamics experiments for them: it melts a crystal, grows one back from a seed, and brackets the melting point, then posts the measured data in the chat as a live chart. Melting and freezing are computed from the forces among atoms. We start with silicon, the most important crystal in the electronics industry. Almost every computer chip, solar cell, and power transistor begins as a single crystal grown from a seed. In the window below, liquid silicon sits on a seed crystal only four atomic layers thick. Press Play and watch the liquid crystallize: atoms build tetrahedra onto the seed one layer at a time, each new layer copying the one below, until everything below the rough top layer is one single crystal. Further down is the same experiment for a metallic, an ionic, and a molecular crystal.
Covalent crystal: silicon. Be patient: the liquid takes ten minutes or more to crystallize on a typical computer.
As the crystal grows, each atom feels only the forces from the atoms around it, computed afresh every femtosecond or two. The ordered lattice, the speed of the growth front, and the temperature at which growth stops all arise from millions of those small calculations. That is what emergent means here: the crystal forms as a result of those forces.
From Melt to Microchip
The chips in a phone, a laptop, or a data center all start as a pure single crystal of silicon. Before a single transistor is patterned, silicon refined to electronic grade — better than 99.9999999% pure, fewer than one foreign atom in a billion — is melted in a quartz crucible at 1414 °C (1687 K). In the Czochralski process, a small seed crystal is dipped into the melt and pulled upward slowly, a few centimeters an hour, while seed and crucible rotate. The liquid freezes onto the seed atom by atom, copying its lattice, and grows into an ingot up to 300 mm across and as tall as a person, one continuous crystal from end to end. The ingot is then sliced into thin wafers and polished to a mirror finish, and only then do lithography, doping, and the hundreds of other steps of chip fabrication begin.
A single crystal is needed because a transistor depends on electrons moving through a perfectly regular lattice. Where two crystal grains meet, or where an atom sits out of place, the lattice is broken: electrons scatter, charges get trapped, and dopants spread unevenly, so devices built on a flawed crystal leak, slow down, or fail. Tens of billions of transistors on one chip need the same perfect crystal under every one of them. Growing a pure single crystal is therefore the first step in making a chip once the silicon is refined, and in many ways the most demanding.
The model above shows that first step at the solid–liquid interface, one atomic layer at a time. Students can see how the seed is copied: each arriving atom takes its place from the layer below, so the whole ingot inherits the seed's orientation. They can also see why the temperature at the interface must be controlled. A little below the melting point the crystal grows, and a little above it growth stops and the crystal can melt back. In the real process, the pull rate and the heat flowing through the interface are tuned to hold the interface just below the melting point for days, the whole time an ingot grows.
Why Start from a Seed?
Heat a crystal far enough and it always melts, but a liquid can be supercooled: cooled well below its freezing point without freezing. Unless dust or a container wall offers a template, a crystal must begin from a tiny ordered cluster that forms by chance, and most such clusters fall apart again before they can grow. This is nucleation. In a simulation the wait for a nucleus can run to nanoseconds or longer — far beyond what a browser can compute while a student watches.
The semiconductor industry avoids the wait when it grows silicon ingots: it gives the liquid a seed crystal to grow on, which removes the nucleation barrier. AIMS does the same on a small scale. Each of the four labs in this article holds a few hundred atoms: a crystal seed a few atomic layers thick at the bottom, its lowest layer or two anchored in place (drawn see-through) and the rest free to vibrate, under a freshly melted liquid with an open surface on top. Below the melting point the liquid freezes onto the seed, layer by layer. Above it the liquid stays liquid, and at most a layer or two next to the seed become ordered. Right at the melting point, crystal and liquid are in balance and the growth front advances and retreats with no net progress. That makes seeded growth a direct way to find the melting point, and it works the same way for every kind of crystal.
Running Experiments by Asking
AIMS is an AI-native environment. Setting up a seeded-growth run, pinning a layer of atoms, and measuring crystallinity are the agent's work. A student asks — "can the liquid freeze again?", "what is the melting point of this model?", "what happens if I heat the crystal?" — and the built-in AI agent recognizes the intent, configures the lab, runs the experiment, and posts a live chart of the results in the conversation.
Go to AIMS and open the list of examples, with the lightbulb button on the toolbar or the Examples item in the main menu (☰). Open one of the seeded-growth examples from the list (Materials Science), open the chat, and type the prompts below. The temperatures are for silicon. For the other crystals, use the ones in parentheses (all but dry ice are also in the table that follows).
Be patient. Every picosecond is computed in your browser, roughly a tenth of a picosecond per second of real time on a typical laptop. If you press Play in an example, copper freezes in a few minutes, salt in five or more, and silicon and nitrogen in ten to twenty. A growth run by the agent first melts the crystal and then regrows it, which takes longer than pressing Play: about ten minutes for copper, up to half an hour for silicon. Finding the melting point chains three or four such runs, from about half an hour for copper to well over an hour for silicon. Leave the tab open and in front while it runs, because browsers slow down pages in background tabs.
Pick a crystal
- Covalent — Silicon Growth from a Seed (Covalent Crystals)
- Metallic — Copper Growth from a Seed (Metallic Crystals)
- Ionic — Salt Growth from a Seed (Ionic Crystals)
- Molecular — Nitrogen Growth from a Seed or Dry Ice Growth from a Seed (Molecular Crystals)
Investigate melting and freezing
- Freezing (growth below the melting point) — “Grow the crystal from a seed at 1500 K.” (copper 1100 K, salt 1000 K, nitrogen 35 K, dry ice 120 K)
- Staying liquid (a run above the melting point) — “Now try it at 1700 K.” (copper 1300 K, salt 1200 K, nitrogen 45 K, dry ice 180 K)
- The melting point — “Find the melting point of this model.”
- Model vs. reality — “How does this model's melting point compare with the real material, and why are they different?”
Each run restores the lab when it finishes, and a "Show final state" button on the chart brings back its last frame. The "Melting of Copper", "Melting of Salt", "Melting of Silicon", and "Melting of Ice" examples start from a solid, for the heating experiments described below.
Four Types of Crystalline Solids
The table summarizes what the four labs show. Each "grows" and "stays liquid" temperature is a measured run, and the model's melting point lies between them. "Time to freeze" is how long the melt takes to become crystal at the "grows" temperature, in simulated time.
| Crystal | Type | Held together by | Seeded growth run | Melting point | Time to freeze | ||
|---|---|---|---|---|---|---|---|
| Grows at | Stays liquid at | Model Tm | Real Tm | ||||
| SiliconSi | Covalent | Shared electron pairs, each atom to four nearest neighbors | 1500 K | 1700 K | 1600–1700 K | 1687 K | ~60 ps |
| CopperCu | A sea of delocalized electrons | 1100 K | 1300 K | ≈ 1200 K | 1358 K | ~20–30 ps | |
| Rock saltNaCl | Ionic | Attraction between Na+ and Cl− ions | 1000 K | 1200 K | 1000–1200 K | 1074 K | ~15–25 ps |
| NitrogenN2 | Molecular | Weak van der Waals forces between N2 molecules | 35 K | 45 K | 40–45 K | 63 K | ~60–80 ps |
Covalent: silicon. In a diamond-structure crystal each atom is bonded to exactly four nearest neighbors at the corners of a tetrahedron. An atom joining the crystal must find four partners at the right angles at once, so silicon grows slowly: about 60 picoseconds at 1500 K, ten minutes or more of watching in a browser. The topmost layer stays rough, because a free silicon surface stays disordered at this temperature. A run at 1600 K still grows, more slowly, and at 1700 K the melt stays liquid, so the model's melting point is near 1650 K — close to the real 1687 K, as the potential's authors intended. The agreement matters because an ingot grower controls how far below the melting point the crystal grows. Here students can see the same melt crystallize at one temperature and stay liquid at another.
Metallic crystal: copper. The quickest to watch: the liquid crystallizes in a few minutes on a typical computer.
Metallic: copper. In a metal, the outer electrons are shared by the whole crystal, and the bond between two atoms depends little on direction. An atom arriving at the growth front only has to settle into a hollow between atoms of the crystal, so copper freezes quickly: at 1100 K the melt is crystal in 20 to 30 picoseconds, two to three times faster than silicon. At 1300 K a layer or two next to the seed become ordered, but the rest stays liquid, and at 1200 K the growth front advances and retreats without finishing. The model's melting point lies near 1200 K.
Ionic crystal: rock salt. Be patient: the liquid takes five minutes or more to crystallize on a typical computer.
Ionic: rock salt. Sodium chloride has a requirement the metal does not: two kinds of ions must each find the right place, every sodium surrounded by six chlorides and every chloride by six sodiums. An ion that lands next to one of its own kind is pushed out again. Even so, salt freezes about as fast as copper: 15 to 25 picoseconds at 1000 K. It takes longer to watch because the long reach of the electric force between ions makes each step more costly to compute. Watch the melt, too: molten salt is much less dense than the crystal, so the liquid starts out swollen and shrinks back as it freezes. At 1200 K the melt stays liquid, which puts the model's melting point between 1000 and 1200 K, bracketing the real value of 1074 K.
Molecular crystal: solid nitrogen. Be patient: the liquid takes ten minutes or more to crystallize on a typical computer, and an occasional run stalls before it finishes.
Molecular: nitrogen. Many solids around us — ice, sugar, wax, most drugs in a pharmacy — are built from molecules held together by forces far weaker than a chemical bond. Solid nitrogen is one of the simplest: N2 dumbbells stacked on the same face-centered cubic lattice as copper, held there by van der Waals attraction so weak that the crystal melts at 63 K. Here everything happens tens of kelvin above absolute zero: the liquid freezes onto the seed at 35 K over 60 to 80 picoseconds, and stays liquid at 45 K. In the grown crystal, the molecular centers settle onto lattice sites, but the dumbbells keep tumbling. That is how many molecular crystals behave near their melting points — the positions freeze first, and the orientations order only at much lower temperatures. (Real nitrogen tumbles on a hexagonal lattice, and this model keeps the cubic one.) Because the molecules are held so weakly and the lab contains only a couple of hundred of them, chance plays a large part. The growth front advances irregularly, and now and then a run stalls halfway. A second run from Reset takes a different path. The Dry Ice Growth from a Seed example shows a harder case: carbon dioxide molecules must also line up in a fixed orientation, and they do.
Read down the "Model Tm" column and the lesson of the four families appears in measured numbers: the melting point follows the strength of what holds the crystal together, from a few tens of kelvin for molecules held by van der Waals forces to a thousand or more for metals, ionic crystals, and covalent networks. The "Time to freeze" column shows something textbooks rarely mention: how fast a crystal grows depends on more than how strongly it is bound. Silicon is slow because each arriving atom must find four nearest neighbors at the right angles. Nitrogen is slow for a different reason: at 35 K its molecules move about four times more slowly than copper atoms do at 1100 K.
Models Make Predictions
Each of the four simulations uses a published interatomic potential — an embedded-atom potential for copper, Tosi–Fumi for salt, Stillinger–Weber for silicon, and a two-site potential for nitrogen. Most were fitted to properties of the cold crystal, such as its spacing, stiffness, and cohesive energy (for nitrogen, rescaled to match the last). Stillinger–Weber's authors also tuned it to silicon's melting point and liquid structure. For the others the melting point is a genuine prediction: salt comes out close to experiment, and copper and nitrogen come out lower, probably because their potentials capture the solid better than the liquid.
Changing the potential. If the melting point comes from the forces, then changing the forces should change it. Ask the agent in one sentence to switch copper to a different published potential and run the same experiment again. With the same atoms, the same lab, and the same procedure, the crystal now melts about 300 kelvin lower than before. Students can then ask the agent why the two potentials disagree.
Other Ways to Study Melting
Watching the order collapse. Open one of the "Melting of" examples (copper, salt, silicon, or ice) and ask the agent to "show how the lattice order changes with temperature." It heats the crystal step by step and measures an order parameter that is high for a lattice and near zero for a liquid, choosing the measure that fits the lattice (rock salt and copper order differently). The curve stays high and then drops steeply. The drop can come well above the melting point. A crystal whose surface does not premelt, as the surfaces of salt and silicon do not, has no place for the liquid to start, so it can be superheated well past its melting point before it melts. Salt stays solid to 1250–1350 K, and silicon past 2000 K. Students can compare the temperature of the drop with the seeded-growth bracket.
The heating curve and latent heat. Ask for "a heating curve" and the agent feeds energy into the crystal at a steady rate and plots temperature against heat added. As the lattice breaks up, the added energy goes into pulling atoms out of their lattice sites, and the temperature stops rising. That energy is the latent heat. In a small simulation the melting step sometimes appears as a flat plateau, and sometimes as a dip: the superheated solid melts all at once, and the latent heat it absorbs comes from the kinetic energy of its atoms, so the temperature falls while heat keeps flowing in. For rock salt the gap between the solid and liquid branches gives a latent heat within a few percent of the handbook value.
From Liquid to Vapor
Melting is half of what a phase diagram shows. A liquid also has a boundary with its vapor, and the Evaporation of Argon example (Chemistry) shows that boundary in a box. The lab holds 192 argon atoms under gravity in a sealed box 30 Å wide and 90 Å tall, at 95 K. The floor is lined with one layer of argon atoms pinned in place and drawn see-through, for the same reason the growth labs anchor their seeds: the walls of the lab only repel, so a liquid on a bare floor forms a drop, and on a floor of its own kind it spreads into a flat pool. Press Play. Atoms leave the pool, fly up through the empty space, strike the walls, and fall back into the liquid. The atoms in flight at any moment are the vapor, and their number settles within a few picoseconds: as many atoms leave the surface as return to it. At 105 K the space above the pool fills with atoms, and at 85 K most of them settle back.
A liquid and its vapor: argon at 95 K. The vapor forms within seconds of Play, and the pinned floor layer is drawn see-through.
The agent turns the picture into a measurement. Ask “How much argon evaporates between 85 and 105 K?” and it holds the pool at five temperatures, counts the atoms in the gas at each, and converts each count to a vapor pressure with the ideal gas law. The count rises from 5 atoms at 85 K to 20 at 105 K, a pressure from 1.6 to 7.2 bar. Plotted against temperature, the points curve upward. Plotted as the logarithm of the pressure against the reciprocal of the temperature, they fall on a line, the Clausius–Clapeyron relation. The slope of that line is the heat of vaporization, 5.2 kJ/mol for this model, and the line reaches 1 bar at 81 K, the model's normal boiling point. Unlike melting, evaporation has no single transition temperature: a liquid is in balance with its vapor at every temperature along the vapor-pressure curve, from the triple point to the critical point, and the boiling point is the temperature at which the vapor pressure reaches the pressure of the air above the liquid, 1 bar. At 2 bar this argon boils at about 89 K.
The handbook values for argon are 6.5 kJ/mol and 87.3 K. The model's atoms attract one another with the measured strength. The lab computes that attraction directly out to 2.5 atomic diameters and adds the rest back as an average pull from the liquid above and below each atom; without that correction the same pool boils at 77 K. The boiling point lands 6 K low, and most of that gap comes from the small pool and the way the gas is counted. The heat of vaporization lands further off. Each point counts between 5 and 20 atoms and scatters by about 15 percent, and five such points fix the slope of the line less well than its height, so the boiling point is the more reliable of the two numbers.
Nitrogen appears in both halves of this article. The molecule that freezes onto a seed at 35 K in the table above also boils, in the Evaporation of Nitrogen example, which holds 144 molecules on a pinned layer of nitrogen. Between 65 and 85 K the agent counts 2 to 10 molecules in the gas, a vapor pressure from 0.6 to 3.7 bar. The fitted line gives a heat of vaporization of 4.0 kJ/mol and reaches 1 bar at 70 K. The handbook values are 5.6 kJ/mol and 77.4 K. This nitrogen is a two-site potential fitted to the real liquid. As with argon, the lab adds back the attraction beyond 2.5 diameters; without that correction the same pool boils at 59 K. The model leaves out the molecule's quadrupole, the small charge pattern that pulls real nitrogen molecules into T-shaped pairs, which moves the boiling point by only about 2 K when it is switched on. Most of the remaining gap comes from the small pool and the way the gas is counted. As with the melting points, the agent reports the model's number next to the measured one and can explain the gap when asked.
Evaporation also cools. The Evaporation of Water example holds 108 water molecules on a pinned layer of water at 395 K, above the normal boiling point. Ask “Open the top of the box and show me evaporative cooling.” The agent switches the heat bath off, removes every molecule that leaves the pool, as a dry wind would, and charts the temperature of the pool against time. The pool cools by about 35 K in 400 picoseconds while 11 molecules escape, about 3 K per molecule, and the cooling slows as the pool cools, because a cooler pool evaporates more slowly. Each molecule that leaves takes its heat of vaporization with it, and the fastest molecules leave first, which is why sweat, a wet cloth, or a puddle feels cold. The heat per molecule comes out near 24 kJ/mol, a rough figure for a film three molecules thick, compared with the handbook value of 40.7 kJ/mol. The pool starts above 373 K and does not boil: with the top open to vacuum there is no pressure for a bubble to push against, and a film this thin has no interior for a bubble to form in, so the water leaves from the surface only.
Open Evaporation of Argon, Evaporation of Nitrogen, or Evaporation of Water from the list of examples (Chemistry) and type the prompts below. A five-point vapor-pressure ladder takes several minutes. The water cooling run takes about ten.
Investigate evaporation
- Vapor pressure — “How much argon evaporates between 85 and 105 K?” (nitrogen: 65 and 85 K)
- The boiling point — “Where does the vapor pressure reach 1 bar, and how does that compare with the real boiling point?”
- Evaporative cooling (Evaporation of Water) — “Open the top of the box and show me evaporative cooling.”
Glass or Crystal: Annealing
A liquid can also become a solid without crystallizing. Cooled fast enough, its atoms are locked in a disordered arrangement, a glass. The Annealing of Amorphous Copper example (Metallic Crystals) opens on such a glass: 108 copper atoms melted at 3000 K and then quenched in half a picosecond. Orbit it and look along any direction: there are no rows. Press Play and, at 150 K, the atoms vibrate, but the arrangement stays disordered, held in a local energy minimum about 90 meV per atom above the crystal.
Annealing moves a glass out of that minimum. Ask the agent "Anneal it: heat to 1000 K, hold for 10 ps, then cool back to 150 K over 20 ps." It drives the whole temperature schedule and charts the energy as it goes. During the hold — below the melting point, so nothing melts — the atoms gain just enough mobility to hop into place, the energy drops in a step, and rows appear. Then press Reset to get the glass back and try two other histories: melt it and quench it to 150 K in half a picosecond, and it is a glass again. Heat it to 1200 K, where the glass melts, and cool it slowly to 150 K over 30 ps, and the liquid crystallizes on the way down. Both runs use the same atoms and end at the same temperature. They differ in the cooling rate, which decides whether a solid ends up a glass or a crystal. These runs are short — tens of picoseconds on 108 atoms — so each takes only a few minutes.
Annealing is also a step in making chips. After a silicon wafer is cut from its ingot, dopant atoms are fired into it by ion implantation, which knocks the lattice near the surface out of order. The wafer is then annealed: heated so the damaged layer can rebuild itself on the perfect crystal underneath and the dopants can settle onto lattice sites. This is seeded growth in the solid state, and it is one of many furnace steps in making a chip.
Why Use AI to Assist Scientific Simulations?
Experiments like these are among the hardest to set up by hand. A trustworthy melting point needs a seed, a melt, a thermostat, and a way to judge crystallinity — and each kind of crystal needs its own force field and its own measure of order. The AI agent handles this setup for the student. Because it maps plain questions onto the right experiment, a student who wonders "why does salt melt at eight hundred degrees but nitrogen at minus two hundred?" can grow and melt both crystals and compare two measured charts.
A melting point, a growth rate, a supercooled liquid, and a glass all emerge from the forces among a few hundred atoms or molecules, computed one femtosecond at a time, while the crystalline fraction and the temperature are measured as they go. The AI agent sets up the simulation that answers a student's question.
What the Simulations Produced
The simulations were given the forces among atoms and molecules. Everything below came out of them:
- Melting points — each model's melting point is measured, and for three of the four crystals it is a genuine prediction.
- Growth by copying — a liquid freezes onto a seed one layer at a time, each layer copying the one below.
- Growth speed — silicon grows two to three times slower than copper, because each arriving atom must find four nearest neighbors at the right angles.
- Swelling and shrinking — molten salt is less dense than its crystal and shrinks back as it freezes.
- Order in stages — nitrogen molecules settle onto lattice sites while they keep tumbling.
- Superheating — salt and silicon stay solid far past their melting points.
- A falling temperature — a heating curve can dip as a superheated solid melts all at once, though heat keeps flowing in.
- Glass or crystal — the same atoms, ending at the same temperature, become one or the other depending only on how fast they cool.
- Vapor pressure — the pressure of the vapor above a liquid rises exponentially with temperature, and the slope of its logarithm against the reciprocal of the temperature gives the heat of vaporization.
- Evaporative cooling — a pool open to vacuum cools as its fastest molecules leave, and cools more slowly as its temperature falls.
Conclusion
This article uses phase changes to illustrate a new way to learn science: students converse with an AI agent that turns a plain-language question into a molecular dynamics experiment and hands back measured data. The four families of crystals — metallic, ionic, covalent, and molecular — become something students investigate: how hot each must be to melt, how fast each can grow, and why. By pairing physically grounded simulations with an AI that can set them up and run them, AIMS turns a student's curiosity into a genuine scientific investigation of how matter changes its state.
← Back to AIMS