Engineering Answers / Production & Quality
Two parts can be the same colour and measure differently, because silicone is translucent, textured and cure-dependent. The tolerance only means something once the measurement conditions are fixed.
For molded silicone, ΔE00 under 1.0 is a tight master match, under 2.0 is a realistic production tolerance, 2–3 is visible side by side, and above 3 is a reject on any cosmetic part. The number is only meaningful with fixed conditions: same illuminant (D65), same observer (10°), same thickness and backing, same surface texture. Change the thickness of a translucent silicone and the reading moves more than the tolerance does.
ΔE is a distance in colour space. The formula has changed twice — ΔEab from 1976, ΔE94, and ΔE00, which is the current recommendation — and the numbers are not comparable between them. Always state which one.
For molded silicone, this is the working scale:
| ΔE00 | What an observer sees | Where it is acceptable |
|---|---|---|
| Under 1.0 | No difference, even side by side | Master standard and first article |
| 1.0–2.0 | Nothing, unless compared directly | Normal production, lot to lot |
| 2.0–3.0 | A slight difference side by side | Non-cosmetic parts only |
| Over 3.0 | Visible without a comparison | Reject on any cosmetic part |
Two rules make the number usable. Specify the formula — "ΔE under 2" without a formula is ambiguous by a factor of nearly two. And specify where the measurement happens: on the part, on a plaque, or on a chip.
Translucency. This is the big one. Silicone scatters light through its thickness, so the apparent colour depends on how much material the light passes through and what is behind it. A 1 mm wall over a white housing and a 4 mm boss of the same compound are the same formulation and different colours.
The practical answer is to measure on a defined thickness over a defined backing, and to check the part in the assembly it will live in. If the customer will see a 1.5 mm wall against a black housing, that is the condition to approve, not a 4 mm plaque on a white card.
Texture. A matte or grained surface scatters light differently from a polished one and reads lighter and less saturated. Two parts from the same cavity, one polished and one textured, will not match on a spectrophotometer even though the pigment is identical. Colour approval has to be against a sample with the final surface finish.
Cure system. Peroxide cure discolours. The decomposition products that have to be baked out are yellow-brown, and a pale or translucent part on a peroxide compound will drift toward cream no matter how the pigment is adjusted. For white, pastel and clear parts, the cure system decides the achievable colour before the pigment does.
Heat and time. Silicone yellow shifts with prolonged heat exposure, and a part that matches at first article can move out of tolerance after an ageing cycle. If the application runs hot or the part is expected to stay white for years, the colour specification should include an aged reading.
Pigment is added as a masterbatch at a let-down ratio, typically a few percent. Two things go wrong:
"Where will this be looked at, against what, and in what light?" We ask it before matching. A customer who approves a colour on a white card under office lighting and then inspects the part in a shop under sodium lamps has not changed their mind — they have changed the measurement condition. Fixing the condition in the specification is cheaper than re-colouring the compound.
A colour callout that can be enforced has four parts:
That last line matters more than it sounds. Two spectrophotometers of different geometry — 45/0 versus integrating sphere — can differ by more than the tolerance on a translucent sample. If supplier and customer are using different instrument types, the argument is about the instrument, not the part.
On an opaque, mid-tone colour in platinum cure, yes, lot to lot with a controlled pigment let-down. On a pale, translucent or bright colour it is difficult, because a small change in pigment loading or cure shows up more against a light background. We usually quote ΔE00 1.0 as a master match, meaning the first article against the approved standard, and 2.0 as the production limit between lots.
Because silicone is translucent. Light enters the part, scatters, and comes back out carrying the colour of whatever is behind it. A 1 mm section over a white backing reads lighter than a 4 mm section of the same compound, and a part over a dark housing reads darker than the same part on a white card. This is not a pigment problem; it is an optical one, and the only fix is to measure under the conditions the customer will see it in.
Yes, and it is the constraint most often discovered late. Peroxide cure leaves decomposition fragments that yellow the compound, which is why near-white and pastel parts are effectively platinum-cure work. Some pigments also interfere with the platinum catalyst, so a colour that works in one cure system may inhibit in the other. Both issues are settled at the compound stage, not at the press.