Chris Collins

Tools · fabrication

Glaze chemistry calculator

Glaze recipe ↔ unity molecular (Seger) formula with a cited materials database — Si:Al, flux balance, the Stull chart, Appen thermal expansion for crazing checks, cone 6 and cone 10 limit formulas, a target formula solved back into a recipe, batches, and line and triaxial blends.

Added

How to use it

A glaze recipe is a list of materials; what melts in the kiln is the oxides inside them. This tool turns a recipe into its unity molecular formula (UMF, or Seger formula) — the moles of every oxide with the fluxes scaled to add up to 1 — and back again, so you can compare glazes, see why one is matte and another glossy, and adjust a recipe by chemistry rather than by guesswork.

Calculation, not a safety test. A formula inside the guidelines does not make a glaze food-safe; only a lab leach test of your glaze, on your clay, from your kiln can say that. No lead materials are included, and you should never use them. Wear a P100 / FFP3 respirator when handling dry materials: silica dust causes silicosis.

How to use it

  1. Recipe → UMF. Pick materials and amounts (or type the recipe as text: custer:20,epk:20,rio:2`; a leading ` marks an addition — colourants and opacifiers counted over the 100 % base). The tool shows:

    • the UMF in three columns: fluxes (R₂O and RO: Li₂O, Na₂O, K₂O, MgO, CaO, SrO, BaO, ZnO) = 1, stabilisers (Al₂O₃ and B₂O₃), glass former (SiO₂), and other oxides and colourants;

    • the Si:Al ratio (SiO₂ ÷ Al₂O₃ — roughly, below 5 tends to matte, 7–10 and above to gloss at cone 6–10), R₂O : RO (alkali : alkaline-earth flux balance), the loss on ignition and the formula weight;

    • the fired oxide analysis in weight % and mole %.

  2. Limit formula. Compare the UMF with a published guideline set: a typical cone 6 limit formula (Digitalfire glossary, after Hesselberth & Roy), Val Cushing’s cone 5–6 and 9–10 glossy, satin and matte limits, and Hesselberth & Roy’s cone 9–10 limits (as tabulated in the Glazy documentation). A bar shows where each oxide sits in its range. Limits are guides: many good glazes sit outside them, and being inside them proves nothing about durability.

  3. Glaze fit. The thermal expansion is estimated with Appen’s additive factors (below). A glaze that expands more than the clay crazes (fine cracks: weaker, and a home for bacteria); one that expands much less can shiver (flake off edges). Calculated values are only meaningful relative to each other — the best check is against a glaze that you know fits your clay: enter it as recipe B and compare. You can also enter a body figure calculated the same way.

  4. Stull chart plots recipes A, B and C by SiO₂ (x) and Al₂O₃ (y) on R. T. Stull’s 1912 map of bright, semi-matte, matte, unfused and underfired glazes, with the crazing zone hatched and the “Q line” of best gloss. Press ⛶ Full screen (or F) for a large chart; Esc leaves.

  5. Target → recipe. Type a target UMF (or copy recipe A’s) and tick the materials you have. The tool finds the recipe with non-negative amounts whose UMF comes closest — Lawson–Hanson non-negative least squares, with each oxide’s error weighted by 1 ÷ max(target, 0.05) so a 0.05 of MgO counts as much as 3 of SiO₂. It reports the formula it actually reached; when the materials cannot supply an oxide (boron with no frit, say) the mismatch shows in the table.

  6. Batch scales the recipe to a batch weight in grams or ounces/pounds, with a running total for weighing into one bucket.

  7. Line & triaxial blends between recipes A, B and C: a line blend of 2–21 tiles from A to B, and a triaxial grid with n divisions per side ((n + 1)(n + 2) ÷ 2 tiles). Each tile shows its recipe, Si:Al, expansion and Stull region.

  8. Materials lists the bundled analyses with their sources, and lets you add your own (from your supplier’s data sheet). Your materials are kept in this browser only and never go into links.

The calculation

For each material, grams × oxide weight % ÷ 100 ÷ oxide molar mass gives moles of each oxide; summing over the recipe and dividing by the total moles of flux gives the UMF. Loss on ignition (carbonates' CO₂, clays' water) leaves in the kiln and only counts in the LOI figure. Molar masses use IUPAC standard atomic weights.

Thermal expansion (Appen). α = Σ xᵢ · aᵢ × 10⁻⁶ /K for 20–400 °C, where xᵢ is the oxide’s mole fraction. Factors (A. A. Appen, as tabulated by A. Fluegel, glassproperties.com): Li₂O 27, Na₂O 39.5, K₂O 46.5, MgO 6.0, CaO 13, SrO 16, BaO 20, ZnO 5.0, Al₂O₃ −3.0; SiO₂ 3.8 below 67 mol %, else 10.5 − 0.1 × mol %; B₂O₃ −1.25 Ψ for Ψ ≤ 4, else −5.0, with Ψ simplified here to (fluxes − Al₂O₃) ÷ B₂O₃ in moles. Oxides with no Appen factor here (TiO₂, ZrO₂, SnO₂, P₂O₅, colourants) count as zero and their share is shown. Checks: pure silica gives 0.5 (fused silica measures about 0.55) and soda-lime float glass about 9.3 (measured about 9). Digitalfire’s Insight uses its own coefficients, so its numbers are on a different scale — compare like with like.

Stull regions are after Stull, “Influences of variable silica and alumina on porcelain glazes of constant RO” (Transactions of the American Ceramic Society 14, 1912), drawn from cone 11 porcelain glazes with RO = 0.3 K₂O · 0.7 CaO, as redrawn for Glazy by Derek Au. Boron, other fluxes, lower cones and slow cooling all move the boundaries.

Checked against

  • Every bundled material’s UMF matches the formula column on its Digitalfire page (Ferro Frit 3134: 0.68 CaO · 0.32 Na₂O · 0.04 Al₂O₃ · 0.64 B₂O₃ · 1.48 SiO₂; Ferro 3124, nepheline syenite, G-200, Gerstley Borate likewise).

  • Digitalfire’s G2926B cone 6 transparent (nepheline syenite 18.3, Ferro Frit 3134 25.4, EPK 19.6, wollastonite 6.9, silica 37.6, talc 2.3) gives SiO₂ 4.94, Al₂O₃ 0.474, B₂O₃ 0.332, CaO 0.602, Na₂O 0.283, MgO 0.077, K₂O 0.039, Si:Al 10.42 and LOI 2.85 % — the same as Glazy’s recipe 6235.

  • Leach 4321 (potash feldspar 40, silica 30, whiting 20, kaolin 10) gives 0.264 K₂O · 0.736 CaO · 0.409 Al₂O₃ · 3.71 SiO₂, as Glazy’s recipe 240.

  • The solver recovers a recipe from its own UMF to within 0.05 %, and reproduces the textbook batch for 0.3 K₂O · 0.7 CaO · 0.4 Al₂O₃ · 3.0 SiO₂ from potash feldspar, whiting, kaolin and silica.

Safety notes

  • Lead: never. It is not in the database.

  • Barium carbonate is toxic; barium glazes that are not well balanced leach barium. Keep them off food surfaces unless lab-tested. Strontium carbonate is the usual safer substitute.

  • Lithium carbonate is toxic if swallowed; manganese fumes and dust are neurotoxic; cobalt and chrome compounds are toxic; copper increases leaching. The tool flags these when they are used.

  • High boron, high alkali and low silica make softer glasses that are more likely to leach; the tool flags those too. Durability also depends on melting fully, the firing and the cooling.

  • Leaching is tested by a lab (for example to ASTM C738 / ISO 6486 for lead and cadmium, or a multi-element test), not by a calculator or a lemon slice.

Limits

  • Material analyses are typical published figures; your bag differs. Use your supplier’s data sheet for real work (Materials tab).

  • The UMF says nothing about particle size, how well the glaze melts at your cone, cooling rate, application thickness or atmosphere — the things that decide whether a glaze is glossy, matte or crystalline in practice. Test tiles decide.

  • Appen’s factors are for glasses, not glazes with undissolved crystals; the expansion is an estimate to compare recipes, not a dilatometer reading.

  • The Stull chart and limit formulas are guides from particular studies, not boundaries.

  • The solver matches oxides, not properties: two recipes with the same UMF can behave differently (raw clay vs calcined, frit vs raw boron).

Related: kiln and firing (cone chart, firing schedules, slip specific gravity), mold maker (plaster slip-casting moulds), chemistry (molar mass and formulas), thermal expansion and heat, recipe scaler.

Sources: Digitalfire Reference Library (Tony Hansen), digitalfire.com — material analyses, the cone 6 limit formula, G2926B, G1214M and G2934; Glazy (glazy.org) and its public data export (github.com/derekphilipau/glazy-data) for the reference UMFs; A. Fluegel, glassproperties.com, for Appen’s factors; R. T. Stull (1912); Glazy documentation, Limit Formulas (Val Cushing; Hesselberth & Roy).

Linking to an example

Share link copies a URL with your recipes and settings — never your own materials. A link can also be written by hand, for example ?r=custer:40. Material ids are listed on the Materials tab.

Parameter Meaning

r, rb, rc

Recipes A, B and C: id:amount pairs separated by commas; +id:amount for an addition

lim

Limit formula: df6, cushing6g, cushing6s, cushing6m, hr10, cushing10g, cushing10m

body

Clay body expansion, ×10⁻⁶ /K

g

Batch size in grams

t

Target formula, e.g. K2O:0.3,CaO:0.7,Al2O3:0.4,SiO2:3

mats

Materials the solver may use, e.g. custer,whiting,epk,silica

nline, ntri

Line blend tiles; triaxial divisions

units

metric (grams) or imperial (ounces and pounds)

Presets (?example=<name>):

example= Shows

g2926b

G2926B, Digitalfire’s cone 6 glossy transparent base

g1214m

G1214M, the “20 × 5” cone 6 glossy base

g2934

G2934, a cone 6 MgO matte, on the Stull chart

leach4321

Leach 4321 cone 10 celadon base with 1 % iron

matte-to-gloss

line blend from the G2934 matte to the G2926B gloss

target-cone10

a cone 10 target formula solved into a recipe

Runs entirely in your browser; nothing you type leaves this page.