Molecular crystallization
Supercooled water
Ice lattice growing · AgI remains
Solid O–H bonds · schematic orientations · enlarged AgI surface
Drag either molecular sample to rotate both views together. Scroll or pinch to zoom, or use the + and − buttons. With a sample focused, use arrow keys to rotate, + and − to zoom, and Home to reset the view.
Start at −12 °C and move model time from 0 toward 10. The close-up shows an example droplet in each population; switch to Droplet comparison to see all 48 droplets on each side. The right-hand close-up shows a reached droplet whenever AgI exposure is greater than zero.
Set AgI exposure to zero to make the two populations identical. Set the temperature above 0 °C to see liquid water on both sides. Around −38 °C, homogeneous nucleation becomes prominent in this illustrative pure-water control.
Changing temperature recalculates the whole exposure at that fixed temperature. It is not a continuous cooling or warming history. Play advances model time; the time slider lets you inspect and replay any stage.
Physical mechanism. AgI can provide a surface that helps an ice nucleus form, reducing the nucleation barrier under suitable supercooled conditions. Water molecules can order near an effective surface and ice can grow from that nucleus. Molecular simulations show that different AgI surface terminations behave differently; a structural resemblance to ice alone is an incomplete explanation. Fraux & Doye (2014), Journal of Chemical Physics, 141, 216101.
What the AgI control means. Each reached droplet is assigned one immersed model AgI particle of identical activity. This controls access to a nucleating surface, not atmospheric concentration, mass dosage or collision efficiency. AgI is retained beneath the ice, and does not supply cooling. AgI nucleation efficiency depends on the particle and experimental conditions. Laboratory work distinguishes immersion, collision and surface-position effects. Nagare et al. (2016), Atmospheric Chemistry and Physics, 16, 8899–8914.
Idealized control. Both populations use matched, equal droplets with no natural ice-nucleating particles, solutes, pre-existing ice or container-wall effects. The model holds temperature constant with an ideal heat sink; it does not calculate latent-heat warming. A natural cloud usually contains other ice nuclei and many droplet sizes. The displayed ice counts cannot be interpreted as cloud-seeding effectiveness.
Illustrative stochastic kinetics. The model assigns reproducible exponential waiting-time thresholds to droplets. Reached droplets have competing homogeneous and AgI-assisted pathways; the earlier event forms ice. The probabilities increase smoothly with cooling. The transition locations and rates are chosen to teach the mechanism and are not fitted to measured AgI activation spectra.
For temperature T in °C and S(x) = 1 / (1 + exp(−x)), the illustrative rates per arbitrary time unit are λₕ = 1.15 S((−T − 37.5) / 0.65) and λAgI = 0.55 S((−T − 8.5) / 2) min(1, (−T / 4)³) for T < 0. Both are zero at and above 0 °C. Event time is the exponential threshold divided by the corresponding rate. These formulas specify this teaching model, not a physical parameterization.
Geometry and growth. The close-up shows complete H₂O units on an ideal tetrahedral hexagonal ice scaffold. Each molecule contains one larger oxygen and two smaller hydrogens in a schematic bent geometry (H–O–H ≈ 104.5°). Solid lines show covalent O–H bonds; dashed links mark the intermolecular ice-network scaffold. Hydrogen orientations are illustrative and do not encode a proton-ordering calculation. Surface reconstruction, charge compensation and molecular forces are omitted. The AgI surface is enlarged and simplified. Growth is deliberately slowed over 3.8 arbitrary time units and does not accelerate when more droplets are exposed to AgI. The paths are animated illustrations, not molecular-dynamics results. Ice-bearing droplet symbols indicate ice presence, not actual snowflake shape.
Scope. This model covers immersed-particle nucleation and subsequent ordering. It does not simulate vapor deposition, contact nucleation, riming, aggregation, secondary ice, precipitation or hail evolution.