# Periodic Table

Version 0.7.1 · 18 November 2026.

Choose an element to load 128 atoms with a 293.15 K kinetic start. Atom, Electron configuration, Crucible, photographs and reference properties follow the selection. The sample starts paused in fixed walls with no background gas. Its initial velocities do not establish thermal equilibrium. Copper uses a published EAM potential; other material samples use labelled cohesive or molecular references.

## Read the state banner

The banner above the crucible describes the **model sample**. Solid-like means a condensed cluster retains its prepared solid arrangement. Gas-like means no atoms meet the condensed-contact criterion. Mixed structure means clustered and dispersed material coexist. Condensed means the available evidence does not distinguish solid from liquid.

A freshly prepared mercury or bromine liquid can show Liquid-like at time zero. After it moves, a condensed reading can replace that preparation label; this is a change in what the evidence establishes, not a freezing event. Generic packing for an unsupported solid form does not become a liquid merely because its internal preparation name is liquid.

Open **Details** for the captured time, structural basis, contact counts and a separate bulk comparison near 1 bar. A hot or brightly lit sample does not automatically become plasma: the crucible has no ion populations, free-electron population or charge dynamics. Plasma is marked **not modelled**. Single atoms, missing observations and imposed gas models retain their explicit limits.

The basic view now keeps the state, temperatures and source controls together. Packing diagnostics, bulk reference points, cross sections and longer methods are in details. Ordinary rates use short decimal numbers; extreme values retain scientific notation.

Electron energy/current edits are drafts until Apply or Enter. A pending note identifies the applied source. View changes and On/Off do not silently replace the draft. Invalid input leaves the source unchanged; an explicit source-type switch or sample reset clears drafts. Sliders commit their own field while retaining a draft in the other field.

Recorded JSON/CSV readings include the state and its basis at capture time. Electron rows retain applied energy/current and leave the incident frequency blank; source power density is jE. Later beam changes do not alter an earlier reading.

## Select groups and periods

Choose Group 1–18 or Period 1–7, or activate a numbered table header. Group, period and search combine; the table dims other elements and List view shows only matches. A filter does not replace the current sample. Select a matching element or press Enter on the search field to load it. Clear removes the filters and query while retaining the sample and view. Header arrows, Home and End move keyboard focus.

The detached rows are in periods 6 and 7. This table leaves their group field unassigned; a Group 3 filter therefore shows Sc and Y. This layout does not settle the competing conventions for the lower part of Group 3.

## Choose a gun

The gun selector switches between photons and electrons in every view. Changing type switches the beam off and prepares a fresh atomic ensemble. Each gun retains its own settings. Turn it on with the shared On/Off control; the material sample remains separate from the isolated-atom reference.

## Use the electron gun

Set kinetic energy from **1 to 3000 eV** and current density from **0 to 10⁶ A/m²**. The current slider has Off and a logarithmic 10⁻⁶–10⁶ A/m² range; exact entry also accepts smaller positive current. Apply commits both numeric fields. Kinetic power per area is `j × E`, with j in A/m² and E in eV.

Hydrogen and helium have selected electron-impact data from [Stone, Kim and Desclaux (NIST, 2002)](https://nvlpubs.nist.gov/nistpubs/jres/107/4/j74sto.pdf). The model interpolates their published cross sections and calculates the ground-manifold excitation coefficient as `(j/e) σ(E)`. Hydrogen uses an explicitly unresolved 1s–2p term; helium uses the selected 1s²–1s2p singlet channel. Values between the physical threshold and the first tabulated point remain unavailable rather than being invented. Other elements have no bundled impact cross section.

An exciting collision leaves a slower outgoing projectile: `Eout = Ein − ΔE`, neglecting recoil. The projectile is not captured. Radiative decay emits the selected atomic branch energy, not the projectile's original energy. These H/He branches are ultraviolet, so their on-screen photon colour is a symbol rather than visible gas glow. The ground configuration stays a reference; moving pulses show transitions, not intermediate occupied levels or electron trajectories.

The impact reference advances **1 ms per real second**. Physical populations and integrated event counts use that clock. Paths and residence drawings are stretched separately to show collision, excitation and radiation in order. Strong streams use weighted glyphs; Off and reduced-motion settings retain quantitative readouts. The restricted model omits other excited channels, superelastic collisions, ionisation products, cascades beyond the selected branches, space charge and transport. It does not predict a discharge spectrum.

In Crucible, an ideal electron receiver deposits all admitted kinetic power in the wall. This is a stated boundary assumption, not a stopping-power or backscatter calculation. There is no charge accumulation or secondary-electron model. Incident electrons are excluded from the electromagnetic source plot; Spectrum identifies their kinetic power separately and retains thermal/neutral-LTE light references. Drawn electron density uses a disclosed 1 ms rate window; it is not a tally of arrivals during a picosecond material frame.

Save setup **version 7** includes gun type, independent photon/electron settings and remembered On values. Versions 1–6 still load as photon preparations. Required fields and power consistency are checked before the visible experiment changes. Restore returns the previous material trajectory and its gun settings, with a fresh atomic ensemble.

## Use the photon gun

Open `index.html`, or visit `/apps/periodic-table/` on SheedyNotes. Scripts, data and 34 real photographs are bundled. No account or external simulation service is required.

The controls below the visualisation set one beam's **vacuum wavelength** and **irradiance**. The wavelength strip covers **0.001 nm to 1000 m**, from gamma rays to radio. Its bands are conventional boundaries on a logarithmic axis; visible light occupies a small part of it. Exact entry accepts nm, µm, mm or m. Open **Beam settings & model** for a sourced atomic resonance or exact irradiance; these controls work in every view.

The idle wavelength field uses six significant digits. Focusing it reveals and selects the full stored value. An untouched Set, unit conversion or intensity edit retains the exact physical beam centre. Typed wavelength edits stay as a marked draft until Set or Enter; intensity changes do not erase that draft. The beam is not rounded to the short display.

Intensity has an Off position and a logarithmic range from **10⁻⁶ to 10¹⁴ W/m²**, with 100 increments per decade. Imported off-grid values are retained until the beam is edited. The default wavelength is 30 THz infrared, about 10 µm. Neither beam control directly sets sample temperature.

The same beam has two distinct uses. In Crucible, the drawn gun sends energy to an ideal wall receiver. That wall assumes 80% absorption at every wavelength; this is not an element's measured absorption spectrum. In Atom and Electron configuration, a separate isolated-atom ensemble follows selected resonant excitation, spontaneous decay and population in lower excited states. Those events do not change the mechanical sample.

Play advances the mechanical sample and Step advances once. **Reset sample** prepares a new 293.15 K kinetic start and wall, switches the beam off and keeps the chosen wavelength. It starts a new trajectory, without claiming thermal equilibration. The atomic reset separately prepares a new ground-state ensemble and appears in the atomic views.

Restore, in Readings & files, returns the previous material trajectory, beam settings and remembered On intensity. It prepares a fresh ground-state atomic ensemble without recovering an earlier population history. A replacement sample's later beam edits do not change the restored sample's remembered intensity.

Playback uses fixed integration intervals and a bounded clock rather than one interval per screen refresh. Target pace is 0.6 ps/s for condensed references, 6 ps/s for material-gas references and 30 ps/s for atomic placeholders. Slow hardware can fall behind; delayed catch-up work is dropped instead of enlarging force steps. Opening an atomic reference view pauses the material trajectory. Atomic populations use their own labelled clock, separate from Play and the picosecond material trajectory; that clock pauses when the document is hidden.

The readouts give the sample's current kinetic estimate, wall temperature, irradiance, accepted power and the unloaded wall's source-only heating rate. Weak input can make very little change over a few picoseconds. Heating rates differ with chamber area and sample coupling.

The periodic table stays on the left; visualisation and element details are on the right. One page scrolls vertically, keeping diagrams, values and population bars together. Windows below 1060 px use Table, Experiment and Details workspaces. Selecting an element updates all three views and keeps the current view open. Search puts an exact symbol, full name or atomic number first, ahead of partial-name matches; Enter and List view use that same order. Model limits opens Sources and moves keyboard focus there, clear of the sticky navigation. Guide and photograph dialogs announce their names and return focus when closed. Arrow keys move table focus; Enter selects. In the crucible, Space plays or pauses, arrows rotate, +/− zoom and Home resets the camera. Touch dragging rotates the chamber; scrolling works outside it. Hiding a running experiment with workspace navigation or a narrow-window resize pauses it.

## Heat transfer and limits

The photon source heats an ideal grey wall, which exchanges energy with atoms through real wall impacts and local floor contact. A fixed, assumed mounting connects the wall to a 293.15 K support with a 25 ps relaxation time. A physically scaled grey far-field term exchanges radiation with 293.15 K surroundings. Exterior vacuum still means no background gas. These wall models do not set or rescale sample velocities.

Heat off removes source input. A hot mounted wall can then cool, and its contact with the sample transfers energy in either direction. A colder sample can keep warming briefly after heat is off while the wall remains hotter. Support and environmental radiation transfers have separate signed energy ledgers. The mounting dominates cooling on the displayed timescale; radiative loss is not artificially accelerated. Interval source power is summed from the current wall updates, so a previous strong beam does not round a later weak beam out of the delivery reading. Cumulative energy totals still have finite floating-point precision.

An **8000 K numerical source guard** reduces accepted heating while dynamics continue. Beam command, admitted power, absorbed power and reflected power are reported separately. Power withheld by this guard is not counted as reflection; the spectrum retains the original command and requested absorption alongside the admitted beam. It is not a validated material-temperature limit. The extended-range notice identifies classical extrapolation: the model has no ionisation balance, reactive bond breaking or universal phase diagram. A force-table or numerical-domain failure still stops explicitly.

Copper density interpolation preserves nonnegative tabulated density near the cutoff. Wall drift resolves collision time and continues with the outgoing velocity. Preparation retains the relaxed floor contact instead of moving the cluster back onto the hard boundary. None of these changes alters the upstream potential table.

## Light and temperature

Sample temperature is labelled a kinetic estimate. Material/cohesive samples exclude centre-of-mass translation; unknown-form atomic placeholders use their container-frame kinetic estimate. Neither establishes equilibrium. For a coherent monatomic cluster, the light reference additionally removes best-fit rigid rotation. Molecular-gas rotational modes are retained. The spectrum notes identify the temperature used for its reference curves.

Crucible photon radiation paths show the assumed receiver: 80% absorbed, 20% diffuse reflection and no transmission. Nonvisible wavelengths use false colour. Paths, wave spacing, speed and packet counts are compressed. Low power may show no packet markers; its numerical flux is still available. The heating kernel uses mean power, not discrete photon arrival sampling.

Open Spectrum for the 100 nm–3 mm thermal window, visible 380–780 nm window or full EM range. Source watts, surface spectral radiance and optional atomic line watts have separate scales. The table keeps its reading position and keyboard focus while values update.

Cu, Au, Ag and Al have published room-temperature bulk optical fits held fixed as temperature changes. Other thermal surfaces use an assumed grey comparison. Selected neutral atomic lines cover H, He, Ne, Ar, Na, Cu, Li and K. Conditional truncated-LTE powers and colours remain separate references. Molecular samples receive no invented atomic LTE glow. Sample reference light does not remove energy from the mechanical model; the ideal wall's environmental radiation does.

All crucible atoms use the same diagram symbol diameter at a given viewport and zoom, in every element and state. The 132 pm reference used by the camera is a drawing convention, not a measured atom boundary. Element-specific covalent radii remain in Details. Neutral gas symbols are soft, translucent grey spheres; their 10% visibility layer is not measured gas transmission. A condensed surface fades continuously as its local structural weight falls, without resizing its atom symbols.

Molecular centres follow the real simulation positions. Nuclei are drawn around each centre with schematic bond spacing so H₂ and O₂ do not collapse into one dot. Current orientation and relative bond motion are retained; the shortest reference bond is drawn as 2.2 symbol radii. This convention applies across phases and changes no coordinates, bond forces or chamber dimensions. The ruler measures centre positions and chamber distances. Molecular trails follow centres. The separate molecule view lists actual model bond lengths in pm.

Optional neutral-LTE gas colour uses calculated visible power per reference atom before screen exposure. Hue comes from the raw visible line powers, independently of the spectrum plot's exposure floor. Its fixed illustrative scale is 10⁻¹² W per atom; a numerical display cutoff of 10⁻⁵ in strength suppresses invisible residue. The earlier scene-strength division capped 128-atom glow at less than 0.2% opacity even for a saturated reference. Physical line powers, particle dynamics and energy ledgers are unchanged. Neutral excitation can emit below the ionisation threshold; this is not a plasma model. Molecular absorption colours, such as chlorine's yellow-green bulk appearance, require a separate absorption calculation and are not painted onto neutral atom symbols. Identity colours remain an explicitly labelled drawing choice.

The **Bulk ref** label compares the light reference's kinetic estimate with available melting and boiling points near 1 bar; for a coherent monatomic cluster, that estimate also removes rigid rotation. Crossing a point changes that reference label and reports the crossing. The separate structural label continues to describe the actual simulation. A small vacuum cluster can rearrange at another temperature; the app does not force particles into a new phase or impose a bulk latent heat at a tabulated point. Material forms and missing values remain explicit. Carbon and grey arsenic receive no invented normal-pressure liquid range; helium has no normal-pressure solidification point.

Symbol text separately shows room-condition state: solid green, liquid blue, gas red and unknown grey. The energy graph contains 773 selected states for 99 spectra plus 19 ground-only records; first-ionisation references cover 108 elements. These are whole-atom levels, not individual orbital energies or solid-state bands.

## Resonant atomic events

A photon can raise a ground-state atom only when the chosen transition is allowed and the photon energy overlaps it. The driven reference currently covers selected ground resonances of H, He, Li, Ne, Na, Ar, K and Cu. Familiar visible noble-gas lines begin in excited states; they are not drawn as direct absorption from the ground state. Missing transition data stays unavailable. Photons above an available first-ionisation threshold receive no invented neutral excitation cycle.

The rate reference uses published Einstein A values, fine-structure degeneracies and an assumed incoherent 1 GHz Lorentzian beam. It represents atoms at rest with statistical sublevel mixing. It includes stimulated-emission competition and saturation, but does not calculate Doppler, hyperfine, isotope or polarisation dynamics. At extreme intensity, omitted multiphoton and ionisation processes can dominate. The energy check also warns when a photon could ionise an already excited atom, even below the ground-state ionisation threshold; it does not invent an ionisation probability.

The model starts in ground, then evolves the ground, upper and selected lower-state populations. Each completed interval assigns floating-point probability residue to the largest population, retaining smaller excitation and afterglow fractions. This prevents long-session drift from stopping the animation. Rate coefficients and integrated event counts are unchanged; numerical energy corrections remain in the balance residual. The displayed scattering rate is the current upper population times its selected spontaneous decay rate. Copper and helium can accumulate population in lower excited states, reducing further fluorescence. Those states are unpumped sinks in this limited network: later recovery and repumping are not calculated.

The drawing now separates absorption, stimulated emission and spontaneous decay. Each uses its own integrated expected count. Stimulated emission adds light in the driving-field direction and returns population to the driven ground manifold; a spontaneous branch can end in another lower state. The matching-field glyph shows the beam centre as a representative cue. The energy arrow and atomic ledger use the selected transition gap. The assumed finite beam linewidth and coherent phase are not resolved in those symbols.

All three channels use weighted paths with a bounded display cadence. Comparing drawn packet counts does not compare physical fluxes: strong stimulated exchange may represent many events per symbol. Open the clock/count line for the three calculated rates and cumulative counts. No minimum event rate is added to make weak beams visible.

Turning the gun off preserves the remaining excitation. It decays through the selected branches, giving afterglow instead of an instant return to ground. Changing intensity or detuning within the same selected pair retains the ensemble; a new element or upper-state pair starts a clearly identified new ensemble. Detuned readouts and excited-ionisation warnings use the retained upper state, rather than a different nearest reference line. Ground-configuration records remain unchanged references.

The photon atomic display clock represents **10 ns per real second**. The display uses integrated expected event counts, with at most four weighted paths per second for each of absorption, stimulated emission and spontaneous decay. A fractional off-beam tail can be shown with its population weight. Travel and residence drawings are schematic; a ring is a state cue, not an electron trajectory. Population bars and rates supply the quantitative reading. Drawing notes expand from the clock/count line. The atomic energy ledger records stored, absorbed, stimulated and spontaneous energy per reference atom, separately from the MD ledger.

**Beam centre** identifies the incoming wavelength and energy. **Spont. branch** identifies the selected spontaneous photon's wavelength and the upper-minus-lower energy gap. They can differ: copper's green decay ends in an excited lower state. The selected endpoint stays visible with reduced animation. This branch readout is an ensemble reference, not a prediction of the next photon.

Visible colour conversion now uses the tabulated CIE 1931 2° standard observer. It fixes a far-red tail error that could turn a normalised potassium line green. Gamut clipping, brightness normalisation, fixed exposure and assumed surface optics still limit screen colour; nonvisible photon glyphs use labelled false colour. [Colour data and licence](data/cie-sources.json).

## Settings and files

Settings groups global appearance, radiation drawing and animation separately from Crucible lighting, bonds and trails. Photons, Waves, Both and Hidden apply to all three views. Hidden suppresses radiation symbols; it does not switch the source off or change model rates or heat. Reduced motion keeps state/rate readings while removing moving radiation paths. Wave marks are schematic, with no wavelength scale. The dialog also provides Save setup and the guide, and keeps Close reachable on short screens. Long element names wrap clear of Atom controls and the Pin button. Wide comparison tables scroll within their own keyboard-accessible panel.

Readings & files provides recording, CSV, setup save/load and Restore. The account link is in the header, clear of the table and diagrams. Auto uses dark for a confirmed member session and otherwise follows the device. Local Dark/Light choices do not edit the account profile.

Version 6 files store physical irradiance and source frequency, element, drawing choices and up to 500 compact reading summaries. Imports accept versions 1–6 up to 16 MB and rebuild the current standard paused preparation and a new atomic ensemble. Earlier formats retain their fixed 30 THz source. Removed environment and count choices cannot reactivate hidden controls; legacy wall-target files load with the beam off. Loading does not restart a material trajectory or recover an earlier population history.

## Sources and verification

Read [science notes](science-notes.md), [material model notes](material-science-notes.md), [crucible physics sources](data/crucible-physics-sources.json) and [third-party notices](THIRD_PARTY_NOTICES.md). Separate ledgers cover [photon excitation](data/photon-sources.json), [phase comparisons](data/phase-sources.json), [elements](data/sources.json), [atomic energy](data/energy-sources.json), [materials](data/material-sources.json), [optics](data/optical-sources.json), [spectral lines](data/spectrum-sources.json) and [photographs](data/images.json).

From the repository root, run `node --test tests/periodic-*.test.cjs`. Browser checks use the existing Playwright installation and local preview; `PERIODIC_ORIGIN` selects its address. The release audit records the browser suite and exact source tested. These checks establish the stated calculations and behaviour, not experimental validity for every element, wavelength or intensity.

Animation audit in 0.7.1: downward cues use their selected lower energy, including helium branches that leave an excited atom. Editing a source cancels stale in-flight drawings without resetting the physical counts or populations. Weighted charged-particle symbols carry sparse minus signs in Crucible. Full transition configurations remain available in the drawing notes.
