Measuring colour in plastics and agreeing a colour tolerance

A colour approved in the office may not look the same next to another part on a shop shelf. That difference has technical causes, and it can be anticipated.

We explain how a colour is measured and compared, what metamerism is and what to agree before production starts. We write from our factory in Tibi (Alicante), where we manufacture by plastic injection and blow moulding. In the From pellets to colour step on our home page, you can see at which point colour is added in an injection moulding machine.

The colour we perceive depends on the light, the object and the observer, all at once. Because each person’s eyesight and the lighting in each place usually differ, two people may disagree about the colour of the same part^{Source 1}.

An instrument does not depend on the observer’s eyesight. It expresses the difference in numbers and finds differences the eye cannot distinguish^{Source 2}. Even so, colour matches are normally checked by eye^{Source 3}, and in plastics visual assessment is combined with measurement^{Source 4,Source 5}.

The reference is the CIELAB colour space, which the CIE (International Commission on Illumination) defines in ISO/CIE 11664-4 for object colours^{Source 6}. It locates each colour with three coordinates^{Source 7}.

• L*, lightness, which runs from dark to light^{Source 8}.
• a*, the position between green, with negative values, and red, with positive values^{Source 7}.
• b*, the position between blue, with negative values, and yellow, with positive values^{Source 7}.

The same point can also be described in cylindrical coordinates, L*C*h^{Source 9}. Chroma C* is the distance from the colour to the lightness axis, and the higher it is, the more intense the colour looks^{Source 4,Source 8}. Hue h is an angle measured from the +a* axis (0°, red) towards the +b* axis (90°, yellow)^{Source 4,Source 7}.

L*a*b* values only make sense if you know which light they were calculated for. That is what illuminants are for, standardised lights defined numerically by their spectral power distribution^{Source 4,Source 10}.

The observer is also specified, and it represents the colour response of a person with normal vision. The 2° observer applies to fields of view from 1° to 4°, and the 10° observer to larger fields^{Source 13}. In the automotive industry D65 with 10° is the usual choice, and in printing, D50 with 2°^{Source 14}.

The distance between two colours is summed up in a single number, the colour difference or ΔE (delta E). There are several formulas, and they will normally give different figures for the same pair of samples^{Source 5}.

ΔE*ab, from 1976, is the colour difference native to CIELAB, the straight-line distance between two points in that space^{Source 6,Source 16}. One unit carries the same weight in any direction^{Source 7}. But visual tolerance is better represented by an ellipsoid, and a box or a sphere around the colour standard may accept some samples that the eye would not^{Source 16}.

CMC, which is set out in the textile standard ISO 105-J03 and is also used in other industries, adapts the ellipsoid to the colour of the standard, so the same limit works for different colours^{Source 17,Source 16}. Its l and c factors weight lightness and chroma, and for plastics l is set at around 1.4 and c at 1, according to one instrument manufacturer^{Source 7}. The maximum accepted value is called the commercial factor^{Source 18}. CIE94, from the 1995 report CIE 116, corrects for the effect of chroma on how differences are perceived^{Source 19}.

CIEDE2000, or ΔE00, comes from the 2001 report CIE 142 and adds corrections for lightness, chroma, hue and the interaction between chroma and hue^{Source 20}. It is standardised in ISO/CIE 11664-6^{Source 21}, and ASTM D2244 recommends it for differences of 0 to 5 ΔE*ab units and describes it as widely used, including in plastics^{Source 5}.

It is a formula for small differences, and a 2005 study on its implementation found incorrect versions online and proposed 34 test pairs to verify it^{Source 22}. Using the formula, verified against those pairs, we have calculated an example for this article, a terracotta red with L* 50, a* 40 and b* 30, shifted by 2 ΔE*ab units in three directions^{Source 21,Source 22}.

By ΔE*ab, the three samples are the same distance away. By ΔE00, they are not. According to one instrument manufacturer, a ΔE00 tolerance can agree with visual judgement around 98% of the time, while an L*a*b* tolerance agrees around 75% of the time^{Source 8}.

Colour comparison bench. At the back stand two plaques: the standard, a terracotta red with L* 50, a* 40 and b* 30, and the sample. In front lies the a*b* plane with the sample marker, shifted by 2 chroma units, and on the left an L* lightness column. On the plane, a small circle marks the colours at a ΔE*ab of 2, and an oval, much wider along the chroma direction, those at a ΔE00 of 2. The sample is at a ΔE*ab of 2.0 and a ΔE00 of 0.6.

Because of metamerism. Two colours with different spectra can have the same tristimulus values under given conditions^{Source 23}, and those three numbers, X, Y and Z, are what the L*a*b* coordinates are derived from^{Source 8}. Two such parts have different spectral reflectance curves, and they can look the same under one light and different under another^{Source 24}.

It happens when the same colour is achieved with different colorants, and buyers usually require a non-metameric match^{Source 25}. Visual assessment matters even more if the part is not made of the same material as the standard^{Source 3}.

Light booth, grey inside, with two bluish-grey plaques on a stand and, beside it, a spectrophotometer showing the reflectance curves of the two plaques, which differ and cross. Under F11 light, a tri-band fluorescent, the plaques look different, at a ΔE00 of 3.0. Under D65 daylight they give the same L*a*b* values and cannot be told apart.

The figure uses a synthetic metameric pair, a bluish grey calculated with the official CIE data to match exactly under D65 with the 10° observer^{Source 26}. Its ΔE00 is 0.8 under D50, 1.0 under F2, 2.8 under A and 3.0 under F11, and 1.8 under D65 when switching to the 2° observer. The F11 value was chosen to make the effect visible and is not a tolerance.

It is detected by measuring under two or more illuminants with very different spectra, such as D65 and A^{Source 24}. The difference that appears when the light or the observer changes is the metamerism index, and the formula used to calculate it must be stated^{Source 27}. ISO 18314-4 describes how to calculate it for a change of illuminant^{Source 28}, and ASTM D4086 how to detect metamerism by eye^{Source 25}. Because colour-difference formulas are optimised for daylight, under other lights the figure is confirmed by eye^{Source 5}.

A colorimeter filters the light reflected by the sample into three bands, red, green and blue, like the cones of the eye^{Source 1}. It is practical for routine checks, but without the spectral curve it cannot detect metamerism or be used to formulate a colour^{Source 1,Source 29}.

A spectrophotometer measures reflectance at each wavelength, normally from 400 to 700 nm and often at 10 nm intervals^{Source 1}. From that curve it calculates the colour under any illuminant stored in its memory, which is why it detects metamerism^{Source 24}, and benchtop models usually also measure in transmission^{Source 1}.

The geometry is the way the instrument illuminates the sample and collects the light, and it changes the measured value^{Source 15}. Here are two of them.

• 45/0, which illuminates at 45° and measures at 0°, or the other way round. It excludes specular reflection and is close to the way the eye sees^{Source 1}.
• d/8, which provides diffuse illumination with an integrating sphere and measures at 8°, with the specular component included (SCI) or excluded (SCE)^{Source 1}.

Specular reflection is light that bounces off as it would from a mirror, and light scattered in all directions is diffuse reflection. On a glossy part the specular reflection is strong and the diffuse light weaker, and on a matt part it is the other way round^{Source 30}. That is why, with a similar tint, a glossy opaque plaque looks darker than a matt one^{Source 31}.

SCE mode measures colour the way a person sees it and is used on the production line to check that it matches the standard. SCI mode measures total reflection, which does not depend on the finish if the material is the same, and is used to adjust colour formulations^{Source 30}.

Yes. ASTM D1729 considers standardised visual assessment essential for critical colour matches and recommends it for selecting or approving a colour^{Source 3}. It is carried out in a light booth^{Source 33}, under simulated D65 daylight, which the standard describes as the one recommended by the CIE, and observers must have very good colour vision and be experienced^{Source 3}.

One instrument manufacturer quotes the clause of the standard that, for critical assessment of materials of medium lightness, calls for between 1,080 and 1,340 lx at the centre of the viewing area^{Source 33}. The instrumental tolerance should be checked against this assessment^{Source 5}.

Plastic is injection moulded from pellets, small granules, and the usual ways of colouring it include masterbatch, pre-blends and pre-coloured compound^{Source 35}. This is the From pellets to colour step on our home page, where the colours shown are examples.

With masterbatch, dispersion matters, meaning that the pigment is evenly distributed. It is achieved when the concentrate is made, with good mixing, high shear (the molten plastic is kneaded vigorously) and the right temperature^{Source 36}.

Even if the colour formulation is correct, the result can vary with the raw material, recycled content, temperature, residence time, moulding, cooling, texture or part geometry^{Source 37}.

• Residence time, the time the pellets spend inside the barrel, exposed to heat and shear^{Source 38}. Process conditions matter more with heat-sensitive pigments, with recycled material and in translucent or thin-walled parts^{Source 37}.
• Thickness. Translucency depends on thickness^{Source 39}, and in transparent film a different thickness changes the colour^{Source 40}.
• Finish. Gloss and texture change the colour you see even when the material is the same^{Source 30}.
• Recycled material. If the input material varies, the masterbatch ratio may need to be adjusted in-line to keep the colour consistent^{Source 41}.

In our guide to types of injection moulds you can see what each runner system involves when you change colour.

ASTM D2244 does not set a figure for plastic parts. According to the standard, the buyer and seller agree on the tolerance and how to calculate it, because each material and each application may need its own, and gloss or texture influence what is acceptable^{Source 5}.

One instrument manufacturer quotes 0.5 CMC or less in the automotive industry and 2.0 or more in printing, and points out that the tolerance should match what the process can reproduce^{Source 14}. These are not figures for your part. They show why the tolerance has to be agreed.

1. Collect parts you would approve and parts you would reject, at least 10 to 20 according to one instrument manufacturer, and measure them^{Source 8}.
2. Set the limit below the ΔE of the closest sample you want to reject^{Source 8}.
3. Check that tolerance against a visual assessment in a light booth^{Source 5}.
4. Ask for ΔL*, Δa* and Δb* as well. ΔE tells you how far the colour is off, but not in which direction, and without the direction you cannot tell how to correct it^{Source 5,Source 9}.

• The colour standard, the approved sample against which production is compared^{Source 7}.
• The formula, with its factors if it has any, and the maximum value^{Source 5}.
• The illuminant, the observer and the lights for the metamerism test^{Source 24}.
• The instrument, its geometry and whether it measures in SCI or SCE^{Source 15,Source 32}.
• How the visual assessment will be carried out^{Source 3}.

If you are preparing a request for a quote, add this information to the part details. In our guide on how to request a plastic injection moulding quote you can see what else is worth sending.

Whatever the customer and the manufacturer agree, depending on the part, its use and what the process can reproduce^{Source 5,Source 14}. Without its formula, illuminant and observer, the figure cannot be interpreted or checked.

Because gloss changes the colour you see^{Source 31}. To compare the colour of the material between two different finishes, both are measured with the specular component included^{Source 32}.

It is more demanding, because the input material varies^{Source 37,Source 41}. A recent study on post-industrial recyclate uses a sensor to measure the colour of the moulded parts, calculates how much masterbatch to add and so achieves more consistent colour^{Source 41}.

You will find more short answers in the frequently asked questions on our home page.

The colour of a part starts to be decided before it is moulded, when you agree how it will be measured and when it will be accepted. At Bersil we manufacture by plastic injection and blow moulding in Tibi (Alicante), with injection moulding machines from 100 to 500 t of clamping force, and less than 1% of our parts are returned to us.

Tell us what colour you need, what it has to match and what light it will be seen under, and we will look into it with you. You can talk to our team, who will give you a 100% human service, or first see what backs up our quality.