A NATIONAL MEASUREMENT A NATIONAL MEASUREMENT

GOOD PRACTICE GUIDE GOOD PRACTICE GUIDE

No. 96 No. 96

Surface Colour Surface Colour

Measurement Measurement
The DTI drives our ambition of The DTI drives our ambition of
‘prosperity for all’ by working to ‘prosperity for all’ by working to
create the best environment for create the best environment for
business success in the UK. business success in the UK.
We help people and companies We help people and companies
become more productive by become more productive by
promoting enterprise, innovation promoting enterprise, innovation
and creativity. and creativity.

We champion UK business at home We champion UK business at home

and abroad. We invest heavily in and abroad. We invest heavily in
world-class science and technology. world-class science and technology.
We protect the rights of working We protect the rights of working
people and consumers. And we people and consumers. And we
stand up for fair and open markets stand up for fair and open markets
in the UK, Europe and the world. in the UK, Europe and the world.

This Guide was developed by the National This Guide was developed by the National
Physical Laboratory on behalf of the NMS. Physical Laboratory on behalf of the NMS.
 Measurement Good Practice Guide No.96

Surface Colour Measurements

Dr Peter J Clarke
Quality of Life Division

ABSTRACT
This document is a guide to the measurement of surface colour. It is primarily concerned with
visible wavelengths of light in the range 360 nm to 780 nm. Methods of assessing and
applying instrumental corrections to the underlying reflectance or radiance factor data are
detailed.
 © Crown Copyright 2006
Reproduced by permission of the Controller of HMSO

ISSN 1744-0610

May 2006

National Physical Laboratory

Teddington, Middlesex, United Kingdom, TW11 0LW

Approved on behalf of the Managing Director, NPL

by Nigel Fox, Quality of Life Division
 i

Contents

Introduction ................................................................... 1

Measurement Instrument Considerations ............................... 3

2.1 Specification............................................................................................................ 4
2.2 Geometric aspects.................................................................................................... 4
2.3 Instrument start up................................................................................................... 5
2.4 Linearity of the measurement system...................................................................... 5
2.5 Reproducibility and repeatability ............................................................................ 5
2.6 Wavelength scale..................................................................................................... 6
2.7 Stray light ................................................................................................................ 6
2.8 Bandwidth ............................................................................................................... 6
2.9 Short and long term aging of instrument................................................................. 7
2.10 Instrument calibration set ........................................................................................ 7

Measurement errors .......................................................... 8

3.1 All instruments ........................................................................................................ 9
3.1.1 Incorrect zero level .......................................................................................... 9
3.1.2 Absolute scale of diffuse reflectance and 0°/45° radiance factor ................... 9
3.1.3 Photometric/intensity accuracy: Non-linearity of the photo-detector and
geometry........................................................................................................... 9
3.1.4 Wavelength error ........................................................................................... 10
3.1.5 Bandwidth ...................................................................................................... 11
3.1.6 Thermochromism ........................................................................................... 11
3.1.7 Sample inhomogeneities ................................................................................ 12
3.2 Integrating sphere instruments .............................................................................. 12
3.2.1 Differing properties of white reflectance standards ......................................12
3.2.2 Specular beam exclusion error (gloss trap error) .........................................12
3.2.3 Specular beam weighting error ..................................................................... 13
3.3 Scanning instruments ............................................................................................ 13
3.3.1 Scan speed and time constant ........................................................................13

Calculation of Colorimetric Data ......................................... 15

Uncertainties ................................................................ 17

Cleaning of Surface Colour Standards....................................19

6.1 Matt ceramic tiles and opal glasses ....................................................................... 20
6.2 Glossy ceramic tiles and opal glasses ................................................................... 20
6.3 PTFE based standards ........................................................................................... 21

Further Reading ............................................................. 22

References ................................................................... 24

Acknowledgements ......................................................... 26
 ii

Contact Details ...............................................................28

List of Figures

Figure 1 Spectra of the same sample taken with bandwidths of 5 nm, 10 nm and 20 nm . 11
Figure 2 Reflectance spectra detail obtained with different time constants....................... 14
Introduction
1
2 Good Practice Guide 96 Chapter 1

This guide is concerned with the measurement of surface colour. Surface colour
measurements are of great importance in the production of a very wide range of
manufactured goods. Industry needs to be able to measure colour within the limits of the
human eye, about 0.5 ΔE ab
*
CIELAB units.

This guide tries to highlight ways of improving operating procedures for surface colour
measurements in order to obtain better results from those measurements. It is assumed that
surface colour measurements are to be made using an internationally accepted measurement
geometry, such as specular included, specular excluded or 0°:45°1. Other measurement
geometries or configurations are also being used to understand measurements of special
effect surfaces such as metallics. They too may be affected by many of the subjects
presented in this guide.

Error sources in surface colour measurements were investigated in a research project looking
at international colour scales, entitled “Harmonisation of National Scales of Surface Colour
Measurements". This project was approved and funded within the EU Fourth Framework
Programme under contract SMT4-CT96-2140. From this project correction factors were
derived along with measurement methods that minimised the uncertainty in surface colour
measurements so that 95 % of the measurements made in different laboratories agreed to
within 0.5 ΔE ab* CIELAB units. The project produced a good practice guide to surface colour
measurements. Some of the contents of that guide have been used as a basis for this guide.

All colour measuring instruments need some means of calibration. This guide considers
items that should be contained in a spectrophotometer calibration set, measurement error and
sources of uncertainty.

The guide is intended to be practical, avoiding equations as much as possible, without

looking in detail to the origin of the problems, but giving solution to them.

Colour measurements made using spectrophotometers are calculations derived from the
underlying reflectance or radiance factor data over the spectral range 380 nm to 780 nm. The
corrections applied to the reflectance or radiance factor data to improve the measurement
result can also be derived and applied outside this wavelength range.

Reporting of colorimetric data also needs to be made in a standard way. Colorimetric data
are always calculated for a specific illuminant (usually one of the CIE Standard Illuminants A
and D65 or CIE Illuminants, for example, C and D50), a specific observer (either the CIE 2°
or the CIE 10° Standard Observer), and a specific measurement geometry.

1
In 2004 CIE changed the manner is which the various recommended geometries are designated. Thus 0°x:45°
replaces 0°/45° where the x indicates that the illumination and the viewing angles are in the same plane. If the
illuminatuion were annular then the new designation would be 0°a:45°. For further information see CIE:2004
Colorimetry2
Measurement
Instrument 2
Considerations

IN THIS CHAPTER Specification

Geometric aspects

Instrument start-up

Linearity of the measurement system

Reproducibility and repeatability

Wavelength scale

Stray light

Bandwidth

Short and long term aging of instrument

Instrument calibration set

4 Good Practice Guide 96 Chapter 2

Colour measuring instruments come in a variety of sizes and configurations. They may be
large spectrophotometers with bolt-on colour measurement accessories, single or dual
geometry bench-top spectrophotometers, hand-held spectrophotometers or colorimeters.
Instruments can be either single or double beam, with one or more light sources and/or
detectors, have diode-array detectors or separate detectors, scan over an extended range of
wavelengths, or have a fixed wavelength range. Many instruments have their own internal
calibration routines and/or calibration artefacts. It is advisable however, to obtain other
calibration artefacts to independently check the instrument performance or ensure that the
samples or artefacts have been traceably calibrated. Spectrophotometers will give spectral
data, which are then used to calculate the colour data. Colorimeters will give only the colour
data, usually for a restricted range of standard observer and standard illuminant
combinations. The availability of this variety of colour measuring instruments means that all
or some of the following need to be considered.

2.1 Specification

Ideally, the process of assessing a colour measuring instrument starts before an instrument is
purchased. The manufacturer will provide some performance criteria for the instrument.
This may include such things as wavelength range, wavelength intervals, wavelength
accuracy, type of sources providing the illumination, source stability, warm-up time,
measurement repeatability, smallest measurable colour-difference, inter-instrument
agreement, operating temperature, cost etc. Most new instruments also have start-up self-test
routines and calibration routines to ensure the instrument cannot be used unless it is fully
functioning.

The user needs to have a clear idea of what measurements, associated uncertainties,
wavelength ranges, etc. they require. This will ensure that the manufacturer’s specification
for the instrument can be compared to the user’s requirements, and appropriate instruments,
short-listed for consideration of other criteria.

Over the life of the instrument, there is a need to record the environmental conditions of the
instrument, sample compartment, room etc., to identify long term changes that are occurring
to the instrument and to identify possible causes of any anomalous effects that may occur.

2.2 Geometric aspects

It is essential to make sure that the instrument’s measurement geometry conforms to

internationally accepted illumination and observation angles; specular included or excluded,
or 0°x:45°, or 45°x:0°. Normal illumination is understood as the beam impinging the surface
at any angle between 0º and 10º. The specular included and excluded geometries involve
using an integrating sphere to make the measurements. The measurement geometries,
angular tolerances, ratio of sphere surface area to area of sphere ports, maximum angle of
acceptance, etc are defined by the International Commission on Illumination (CIE) and are
given in CIE Publication 15:20042.

Consideration also has to be given to where the light dispersing element is placed in the
system. This is usually a diffraction grating but could be a prism or a series of filters. Light
Chapter 2 Good Practice Guide 96 5

dispersion could happen before the sample, resulting in the sample being illuminated with
only a small band of wavelengths – monochromatic illumination, or after the sample, in
which case the sample is illuminated with the full wavelength range – polychromatic
illumination. This has implications for the type of detectors used in the system. Instruments
with light dispersion before the sample tend to be scanning instruments that have a single
detector output. The detector is sensitive to a range of wavelengths. Instruments with
dispersion after the sample usually have multiple detector outputs, e.g. from a diode-array
mapped to the wavelength range of the instrument. If wide spectrum illumination is used
(incandescence lamp, continuous xenon lamp, xenon flash lamp) be aware of fluorescent
samples. The source has to match a CIE standard or recommended illuminant (A, C, D65
etc.) as closely as possible to correctly determine colour coordinates. If this is not
accomplished, the surface colour of fluorescent samples has to be measured by the double
monochromator method3. Fluorescence is present in many objects and even in some white
standards. Another thing to bear in mind is that heating and thermochromism effects may
give measurement problems when the sample is directly irradiated with the full wavelength
range of the source.

2.3 Instrument start up

Perform the start up routine as indicated by the instrument’s manufacturer. Currently this is
automatically carried out by most spectrophotometers and colorimeters when switched on.

The period of time elapsing for stable results should be determined. Stable results occur
when the change in the instrument reading from one value to the next is within acceptable
limits. The instrument manufacturer may suggest a warm-up time. Generally, let the
instrument warm up for a period of time that has been estimated to give stable
displays/results, but never less than the time suggested by instrument’s manufacturer, or
thirty minutes.

2.4 Linearity of the measurement system

The accuracy of an instrument requires either a linear instrument response or accurate

knowledge of the deviations from linearity. A linear instrument response means that, over a
specified range, the output signal of the detector is exactly proportional to the input signal.
Any non-linearity in the output of the detector system, that is the detector and it’s associated
electronics, will change the measurement results. Checking the linearity is discussed in more
detail in section 4.1.

2.5 Reproducibility and repeatability

Reproducibility data is obtained by measuring a sample, removing the sample, replacing the
sample and making another measurement, and then repeating this sequence a number of
times. The replacement of the sample in the instrument needs to try and achieve the same
alignments and measure the same area of the sample at the same angle as before. The
standard deviation of these measurements divided by the square root of the total number of
6 Good Practice Guide 96 Chapter 2

measurements (the standard error), gives an indication of the reproducibility and can be
included in any total measurement uncertainty assessment.

Repeatability data is obtained by measuring a sample, leaving it in place and making another
measurement, and then repeating this a number of times. The standard deviation of the
measurements divided by the square root of the total number of measurements, then gives an
indication of the repeatability and can be included in any total measurement uncertainty
assessment. This is a Type A (evaluated by statistical methods) uncertainty4.

2.6 Wavelength scale

Wavelength scales in colour measuring instruments are difficult to assess unless you are
using a scanning spectrophotometer, where a narrow emission line discharge lamp or some
other wavelength standard may be used. This is discussed further in section 3.1.4

2.7 Stray light

Stray light is difficult to evaluate. It is light flux reaching the detector outside the spectral
distribution prescribed for the measurement or which reaches the detector by a path other
than that prescribed. Stray radiant energy is both heterochromatic - of different wavelengths
to those set by the spectral bandpass - and isochromatic - of the same wavelengths as those
set by the spectral bandpass - but does not travel via the sample. Stray light can be internal or
external.

External stray light is that from any source other than the instrument source. It is most likely
to be found in instruments which are not adequately enclosed so that light may enter. The
obvious way of checking for external stray light is to darken the room or cover suspect areas
and note if there is any change in the output signal.

In colour measuring instruments, stray light generally should not be a problem if the
instrument designer and manufacturer have done their job properly. Check the
manufacturer’s specification for any assessment of this they may have done.

2.8 Bandwidth

For an instrument where the spectral bandwidth cannot be set by the user, it is important to
know the value used. It is also important to know the form of the slit function. This is the
spectral distribution of the light incident on the sample when a single wavelength is selected
in the spectrophotometer. It is usually triangular in a scanning spectrophotometer, but not in
a non-scanning system, such as those using detector arrays.

Bandwidth is usually not independent of wavelength although it is frequently assumed to be

so.
Chapter 2 Good Practice Guide 96 7

2.9 Short and long term aging of instrument

Once a spectrophotometer has been assessed as passing the required specification and
meeting the user’s requirements, steps need to be taken to ensure that it continues to meet
them during use.

Instruments have to be maintained as indicated in their manual. It is necessary also to pay

special attention to the handling and maintenance of the instrument’s calibration standards.
For those instruments with an integrating sphere, pay attention to the sphere coating.
Degradation of this coating produces several errors.

Simple checks should be done every day or every time the instrument is switched on for use.
This can be as simple as checking the reflectance at a couple of levels. Then on a longer time
scale, e.g. weekly, depending on use and or regulatory requirements, more checks should be
made. For example, check the value of reflectance at several levels or wavelength scale at
several wavelengths. A full check using all standards should be made periodically as well,
e.g. monthly. Using this procedure a history of the instrument’s performance can be built up.
This may aid in narrowing down the time of occurrence of a problem and in diagnosing the
fault. In all cases, if the instrument gives a measurement result outside the value for the
standard, then the cause should be investigated, and a service engineer called if the
instrument is at fault.

It is useful to also record environmental conditions at the time of measurement as they may
influence the measurement result and cause anomalous effects. E.g. temperature of cell,
sample compartment temperature, room temperature and humidity.

2.10 Instrument calibration set

Many instruments are supplied with their own calibration set, which contains at least a white
standard. A black, or zero reflectance, standard along with other colours may also be
included. Often the supplied set may not be comprehensive or have a traceable calibration
and therefore it is advisable to buy another more comprehensive calibration set to
independently check the instrument performance. This set should ideally contain the
following: two white standards (one matt and one glossy), a zero reflectance standard (black
trap for reflectance, black glass for 0°/45° radiance factor), a grey standard with about 50%
reflectance, a mirror (for reflectance) and a wavelength standard. The function of each of
these standards is described in section 3. The standards should be handled carefully to ensure
they are not damaged in use. They should also be kept clean as any dirt, fingerprints, etc on
them will change the calibration values.
Measurement errors
3

IN THIS CHAPTER All Instruments

» Incorrect zero level
» Absolute scale of diffuse reflectance and
0˚/45˚ radiance factor
» Photometric accuracy: Non-linearity of the
photo-detector and geometry
» Wavelength error
» Bandwidth
» Thermochromism
» Sample inhomogeneities

Integrating sphere instruments

» Differing properties of white reflectance
standards
» Specular beam exclusion error (gloss trap
error)
» Specular beam weighting error

Scanning instruments
» Scan speed & time constant
Chapter 3 Good Practice Guide 96 9

Spectrophotometric errors are discussed in this section on a qualitative basis. All are known
to affect the measurement result, although by different degrees for different instruments. It is
the user’s responsibility to check the instrument or to trust the manufacturer’s specification in
this sense.

Not all of the errors affect all instruments. Some, such as the gloss trap error and the specular
beam weighting error, are specific to integrating sphere instruments. Others, such as
scanning speed and integration time, are specific to wavelength scanning instruments.

3.1 All instruments

3.1.1 Incorrect zero level

Detector noise and stray light within the instrument optics will produce a signal for zero
reflectance or zero 0°x:45° radiance factor. This value must be accurately determined and
then subtracted from all subsequent readings. It is possible that the dark reading changes
with wavelength, so if possible spectral data should be obtained for a zero reflectance or
radiance factor standard at the sample position. Simply making a spectral scan with the
sample port open is not satisfactory. It is better practice to use a black glass wedge trap at the
sample port for the zero reflectance spectral scan for integrating sphere instruments, and a
black glass plate for 0°x:45° radiance factor instruments. Some commercial gloss traps
provided with integrating sphere instruments to exclude the specular component are not
satisfactory as zero reflectance devices. This wedge should not reflect any significant part of
the incident light. This will occur if the specular beam undergoes three or more specular
reflections within the wedge. As a good rule of thumb, a wedge whose end cannot be seen
under standard laboratory illumination is usually a good zero reflectance specimen.

3.1.2 Absolute scale of diffuse reflectance and 0°/45° radiance factor

The instrument scale should be calibrated against standards traceable to a national metrology
institute. The calibration tiles provided with many instruments may be traceable to different
national scales which may have biases with respect to each other. If measurements made on
different instruments are to be compared, then these biases need to be removed and
instruments ideally calibrated with white standards that are traceable to the same scale. It
needs to be noted that standards used to calibrate instruments have an inherent uncertainty
derived from their own calibration with respect to any national standards laboratory’s
reflectance and radiance factor scales. Also, if the standards are not properly maintained and
recalibrated at intervals appropriate to their use, the calibration factors will not be correct,
leading to erroneous measurements. Differences of 0.5% in absolute value of radiance factor
measurement will produce measurable errors in terms of CIELAB units.

3.1.3 Photometric/intensity accuracy: Non-linearity of the photo-detector and geometry

Photometric accuracy is mainly a combination of two factors. One is related to the
geometrical characteristics of the beam used in the spectrophotometer and the other is due to
the linearity of the instrument’s photodetection system. From optics, it is known that
reflectance depends on the angles of incidence, numerical aperture, spectral content and
polarisation of the incident beam, together with the angles of collection, numerical aperture,
spectral content and polarisation of the reflected beam. Most spectrophotometers have beams
of low divergence, but not exactly parallel, very close to normal incidence, polarised to some
extent and differ from instrument to instrument. Therefore, even when no error would affect
10 Good Practice Guide 96 Chapter 3

the measurement, the results may differ from instrument to instrument if the beams are not
exactly identical. Fortunately, this contribution to photometric accuracy is small in a high
quality instrument and the largest contribution is that due to lack of linearity in the detection
system.

The response of a spectrophotometer’s detector to light is not always linear throughout its
entire range. Towards the higher end, saturation of the detector or electronics may occur.
Towards the lower end, the noise floor will limit the minimum measurable quantity. Even in
the middle range some effects may occur that modify the linear behaviour normally expected.
To check for this, a grey tile of approximately 50% reflectance is the minimum needed. A
set of several grey tiles with different reflectance values would give more information about
the linearity. The presence of the tile may also modify the response of the system. The zero
and 100% points can be assumed to be accurate if zero level and scale corrections have been
made. Then a formula, based on a polynomial, can usually be fitted for linearity correction
across the range tested.

3.1.4 Wavelength error

Wavelength errors will lead to colorimetric errors, so the instrument’s wavelength scale
should be checked. Several wavelength standards are available including spectral emission
line sources, reflectance tiles and transmission filters (e.g. McCrone, Holmium). The user
must use the most appropriate for their instrument type.

To calibrate the wavelength scale in a scanning spectrophotometer with a reflection type

wavelength standard, an ordinary reflectance measurement should be made within the
interval in which wavelengths are certified, setting the instrument with the same bandwidth
and a combination of scanning speed and time constant to be used in the measurement. If the
instrument allows for transmission measurements, a transmission filter may be placed at
some point in the sample beam, with a white standard at the sample port. Then the
wavelength can be calibrated using absorption peak filters by measuring their transmittance
with the same parameters as for the measurement.

In a non-scanning instrument checking the wavelength is much more difficult. A reflection

wavelength standard should be placed at the sample position and a measurement made using
the normal procedure. The position of the wavelength peak or feature can be determined
from the spectral data and compared to the calibration position.

If the wavelength error is less than the uncertainty of the standard, no correction should be
applied. If the wavelength error is less than 0.1 nm, no correction should be applied because
the subsequent colorimetric error is negligible. In other cases, reflectance values should be
corrected according to the expression:

∂Rm (λ )
R(λ ) = Rm (λ ) + Δλ (1)
∂λ

where R(λ) is the correct reflectance value, Rm is the measured reflectance value and Δλ is
the wavelength scale error, the difference between the actual and the displayed wavelength.

Clearly, this error cannot be checked or corrected in those instruments in which reflectance
values are not provided to the user.
Chapter 3 Good Practice Guide 96 11

3.1.5 Bandwidth
The spectral bandwidth used in recording the reflectance spectrum has an effect on the
acquired spectral values. Figure 1 shows two spectra of the same sample obtained with
different bandwidths. Clearly this will result in differences between the values of the
associated colour coordinates if no action is taken.

From a practical point of view, the error in colour coordinates is negligible if the reflectance
spectrum does not change significantly within the spectral ranges compared to the bandwidth
value. In most cases a bandwidth up to 5 nm can be used in recording the reflectance or
radiance factor spectrum without altering the spectrum. For many instruments this parameter
cannot be set by the user, although it should at least be known.

Figure 1: Spectra of the same sample taken with bandwidths of 5 nm, 10 nm and 20 nm

Reflectance vs Wavelength
100

80
Reflectance (%)

60

40

20 5 nm data
10 nm data
20 nm data
0
300 350 400 450 500 550 600 650 700 750 800

Wavelength (nm)

This source of error affects the value of colour coordinates very much like a wavelength
error. It is more noticeable in chromatic samples with steep spectral slopes in the reflectance
data.

3.1.6 Thermochromism
Thermochromism is change of colour with changing temperature. When specifying the
colour of a material the measurement temperature should be specified. Generally the more
highly coloured the sample the more thermochromic it is likely to be. This effect is most
prominent in materials having spectral distributions with steep slopes, such as those coloured
red, orange or yellow. A correction method has been published in the literature for ceramic
colour standards5-8.

In relation to this item, it is advisable to check for temperature difference between the sample
placement position and the room temperature. Up to 12 ºC temperature difference has been
found in an instrument, but two or three degrees is acceptable.
12 Good Practice Guide 96 Chapter 3

3.1.7 Sample inhomogeneities

Samples and standards are generally not 100% uniform across their surface8. Therefore, the
measurement result is an averaged value within the area covered by the incident light. To
minimise this error, the reflectance should be measured at least twice by rotating the sample
about the optical axis. The final reflectance value should be the mean of the measurements.

3.2 Integrating sphere instruments

3.2.1 Differing properties of white reflectance standards

It has been experimentally shown that when making measurements of matt samples using
spectrophotometers which include an integrating sphere accessory, errors may be introduced
if the instrument is calibrated using a glossy standard9. The same is true for glossy samples
calibrated using a matt standard. To avoid this effect it is good practice to measure matt
samples against a matt white standard and glossy samples against a glossy white
standard.This error should not affect 0°x: 45° radiance factor measurements unless the
aperture angle of the instrument varies significantly with respect to that used in the
calibration of the standard.

3.2.2 Specular beam exclusion error (gloss trap error)

A gloss trap which does not completely eliminate light specularly reflected from samples can
lead to an error when measuring glossy samples in the specular excluded geometry. The
amount of error depends on the specific instrument. A method to check for the size of this
error consists of measuring the reflectance of a good quality mirror against a matt white
standard with the integrating sphere set for specular excluded geometry10. A correction
coefficient may be calculated from this measurement according to the expression (all
reflectances in %):

ρ m (λ )
K 2 (λ ) = Rt (λ )
4
M (λ ) ρ t (λ )

Where ρm is the instrument’s reading when the mirror is measured, ρt is the reading when the
white standard is measured, M is the mirror’s reflectance, the value 4 stands for the nominal
specular reflectance of a glossy sample, and Rt is the white standard’s actual reflectance (all
reflectances in %). If K2 < ± 0.05 then the gloss trap is performing well. If larger than this
value then the instrument may need realigning. Alternatively, if no linearity correction has
been applied to the measurements then a sphere error correction may be applied. This is
effective when you have no grey standards and have only used a white standard to calibrate
the instrument. The corrected reflectance value would be given by:

ρ s (λ ) ⎛ ρ (λ ) ⎞
R(λ ) = Rt (λ ) + K 2 (λ ) ⎜⎜ β s − α ⎟⎟
ρ t (λ ) ⎝ ρ t (λ ) ⎠

where ρs is the reading when the sample is measured, ρt is the reading when the white
standard is measured, and α and β are coefficients related to the glossiness of the sample and
Chapter 3 Good Practice Guide 96 13

the white standard, respectively, and whose values are 1 or 0 for glossy and matt specimens
respectively.

Clearly this coefficient cannot be determined in those instruments that do not provide
reflectance values, but only colour coordinates. In this case, the measurement could be done
and the coordinate representing lightness (Y or L*) could be examined. If this value is very
small, close to zero, then the gloss trap correction error will be negligible. If this value is not
close to zero, then the error will be significant but no correction can be applied.

3.2.3 Specular beam weighting error

In many integrating spheres, light specularly reflected may not be evaluated with the same
efficiency as light diffusely reflected and an error will be present in the total reflectance
measurement of glossy samples. The size of this error depends on the specific instrument.

Like the specular beam exclusion error, a method to check for the size of this error consists of
measuring the reflectance of a good quality mirror against a matt white standard with the
integrating sphere set for specular included geometry10. A correction coefficient may be
calculated from this measurement according to the expression (all reflectances in %):

⎛ R (λ ) ρ m (λ ) ⎞
K 3 (λ ) = 4 ⎜⎜ t − 1⎟⎟
⎝ M (λ ) ρ t (λ ) ⎠

If K3 < ± 0.05 then the specular beam is being collected with the same efficiency as the
diffused light. If larger than this then the measurements may need correcting. If no linearity
correction has been applied to the measurements then a sphere error correction may be
applied. This is effective when you have no grey standards and have only used a white
standard to calibrate your instrument. The corrected reflectance value would be given by:

ρ s (λ ) ⎛ ρ (λ ) ⎞
R(λ ) = Rt (λ ) + K 3 (λ ) ⎜⎜ β s − α ⎟⎟
ρ t (λ ) ⎝ ρ t (λ ) ⎠

Clearly this coefficient cannot be determined in those instruments that do not provide
reflectance values, but only colour coordinates.

3.3 Scanning instruments

3.3.1 Scan speed and time constant

An incorrect setting of the scan speed and time constant leads to an erroneous recording of
the reflectance spectrum, as shown in Figure 2, and leads to a subsequent error in calculating
the colour coordinates.

Figure 2: Reflectance spectra detail obtained with different time constants

 14 Good Practice Guide 96 Chapter 3

Relative reflectance vs wavelength for different time constants,

t, at a scan speed of 60 nm/min
1.2

1.0
Relative reflectance

0.8

0.6

0.4 t = 0.2 s
t = 0.5 s
t=1s
0.2
t=2s
t=5s
0.0
580 582 584 586 588 590 592 594 596 598 600

Wavelength (nm)

The concept of time constant is used in a scanning spectrophotometer in an ambiguous way,

because it relates to the time needed to give a full response to a steep slope change, but
different manufacturers associate different limits to the proportion of the final response (68
%, 95%, 99%, etc.). It is necessary to carefully look at the specification of this parameter in
the instrument’s manual. A good rule of thumb is to use a time constant that reaches 99% of
the final response after a steep slope change within the time interval used to sample the
reflectance spectrum. If t is the time needed by the instrument to produce 99% of the final
response to a steep slope change and d is the sampling spectral interval, the correct scanning
speed will be given by d/t.

Which parameter (scan speed or time constant) should have preference? It depends on the
specific measurement. Those situations where noise has to be low need the time constant to
be set first and then the scanning speed. Those situations where the measurement time has to
be minimised, need the speed to be set first and then the time constant.

The effect of this error looks like a wavelength error and it is more evident in the highly
chromatic samples and negligible in those that are less chromatic, such as white or grey
samples.

Some current scanning spectrophotometers do not allow the user to set the time constant, just
the scanning speed. Perhaps the reason is to avoid this problem!
Calculation of
Colorimetric Data 4
16 Good Practice Guide 96 Chapter 4

Calculation of colour coordinates from reflectance values has to be carried out by following
the recommendations of international bodies like CIE (Commission International de
l’Eclairage)2. Tabular data for CIE Standards Observers and Illuminants are given for a 1 nm
bandwidth, therefore, from a rigorous point of view, they only could be used with reflectance
spectra obtained using that bandwidth and that sampling interval. From a practical point of
view however, they can be used, as they often are, for spectra obtained with bandwidths up to
5 nm, using a summation step of the same width. If the measuring instrument uses
bandwidths larger than 5 nm to measure the reflectance, then the preferred option is to
interpolate the spectral data to 1 nm or 5 nm interval data, correct to 1 nm or 5 nm bandwidth
respectively and use the appropriate CIE tables in the calculation of colorimetric data.

For smoothly changing spectral profiles with no discontinuities a practical alternative is to

use weighting functions for a given standard illuminant and observer combination, with the
assumption of an ideal band-pass function of specified width at half-height. If the source and
other features of the instrument are well known, deviations from the ideal band-pass function
can be computed. For instance, those functions recommended by ASTM (American Society
for Testing and Materials)11.

Colorimetric calculations are usually carried out within the software of instruments. It is the
user’s responsibility to verify whether they have been done correctly. The amount of the
potential error depends on the sample, but it may be as large as several CIELAB units if a
large spectral bandwidth is used or the sample is of very low reflectance.

Instruments which give only colorimetric data, such as colorimeters, contain filters that have
spectral sensitivities designed to mimic the colour matching functions. There are limitations
on colorimetric accuracy depending on the closeness of match of the filters to the colour
matching functions and the need for the light source(s) to closely match the standard
illuminant that the data are required for. Colorimetric data are usually calculated by the
instrument from the CIE X, Y and Z tristimulus values the filter readings are converted into.
Uncertainties
5
18 Good Practice Guide 96 Chapter 5

A measurement result will be incomplete if it is not accompanied by its uncertainty value.

Currently, there is no universally accepted method for calculating the uncertainties to be
associated with colorimetric coordinates, although some methods have been published10-12. It
is assumed however, that at least the following uncertainty sources should be considered in
the estimation of the final uncertainty. Values assigned to each of them, and the method used
to compute the combined final uncertainty in colour coordinates, should be specified. It is
important to remark that the uncertainty value will not disappear if a correction factor is
applied to the measurement, because the correction factor itself will have an associated
uncertainty.

Uncertainty sources:

Scale uncertainty (usually via reference standard, e.g. white standard)Photometric linearity
uncertainty
Dark level uncertainty
Wavelength uncertainty
Bandwidth uncertainty
Gloss trap or specular beam uncertainty, whichever applies, for integrating sphere
geometries.
Repeatability uncertainty

Where spectral reflectance data are available the most common practice used to calculate
uncertainties in colour coordinates is to add in quadrature the values assigned to the previous
quantities so that a reflectance combined uncertainty value is obtained. This may be a single
value or a set of spectral values. Then there are two options. The first is to apply the partial
derivative rule to the colorimetric tristimulus equations, to obtain uncertainty values for X, Y
and Z and then the partial derivative rule is applied again to the equations defining the colour
coordinates of interest in terms of X, Y and Z to obtain the associated uncertainty (x, y, Y),
(L*, a*, b*), (L*, u*, v*), etc.122

The second uses numerical methods with no need of deriving partial derivatives analytically.
The partial derivatives of a colour quantity in terms of each spectral value are obtained by
simply calculating the colour coordinates with and without a small change of each input
spectral value. The uncertainty of the colour quantity is then obtained as a combined
uncertainty for all spectral values13.

2
See also NPL Best Practice Guide No. X Uncertainties in surface colour measurements
Cleaning of Surface
Colour Standards 6

IN THIS CHAPTER Matt ceramic tiles & opal glasses

Glossy ceramic tiles & opal glasses

PTFE based standards

20 Good Practice Guide 96 Chapter 6

Ceramic tiles and other colour standards may over time become dirty and need cleaning.
Standards with glossy surfaces are easier to clean than standards with matt surfaces. The best
way to avoid having to clean standards is by careful handling of them to minimise build up of
dirt on them. It is important to pay particular attention, to only handle standards by the edges
and not touch the front surface, to keep the standards in a closed box when not in use and to
make sure that any surfaces such as instrument ports they are put up against are themselves
clean.

6.1 Matt ceramic tiles and opal glasses

It is best to avoid getting matt tiles dirty by handling them as little as possible; wear clean
cotton gloves, hold the tiles by the edges and always keep them in a closed box to prevent
accumulation of dust etc. on them when not being used. Matt tiles are abrasive and should
not be put against dark surfaces, e.g. black paint, as the surface might rub off onto the tile.
Measurement ports should be protected with thin white non-fluorescent paper.

If cleaning is necessary, then it should start with the lowest level of cleaning and progress up
a level if the tiles remain dirty. The simplest level is to use filtered (carbon + 22 μm paper)
compressed air to remove dust and particles from the surface of the standard. The
compressed air should be filtered to ensure that no oil that may be in it from the compressing
process reaches the standard.

For heavily soiled matt tiles ethanol or propanol should be applied with a soft brush, and
followed, while still wet, with a dilute non-fluorescent detergent, again applied with a soft
brush. The tile should then be rinsed in warm running tap water for at least 1 minute
followed by running deionised or distilled water for at least 1 minute. The tile should be
shaken and then blown with filtered compressed air to remove excess water and left for at
least 24 hours to dry. If results are unsatisfactory then the above process should be repeated
using a more concentrated detergent solution, followed with washing in warm running tap
water for ten minutes, and then running distilled water for 1 minute, then dry as before. After
this type of heavy cleaning it is recommended that the tile be recalibrated if it is used as a
standard.

6.2 Glossy ceramic tiles and opal glasses

Glossy tiles can be cleaned by firstly using filtered compressed air to remove dust. Secondly
by breathing on them and rubbing in a circular motion with clean lens tissue or a clean soft
tissue. This should be alternated until breathing produces a uniform breath pattern.

If they become very dirty then, before the above procedure is followed, they should be
cleaned first with lens tissue or a soft tissue soaked in ethanol or propanol, followed by
rinsing in deionised or distilled water for 1 minute and drying with a soft tissue.

6.3 PTFE based standards

Chapter 6 Good Practice Guide 96 21

There are a number of PTFE based standards available from various manufacturers. These
included such products as Spectralon®, Fluorilon®, Zenith®, halon etc. As with other
standards the simplest level of cleaning is to use filtered compressed air to remove dust and
particles from the surface of the standard. If a heavier cleaning is needed then the cleaning
procedure is to remove the top surface to expose a fresh clean surface.

For all standards except those with a reflectance of less than 10%, sand with a circular
motion under running water using a 220 – 240 grit emery cloth wrapped round a sanding
block. Sand until the surface is totally hydrophobic (water beads run off immediately). Then
blow dry with filtered compressed air, or allow the material to air dry. For standards with a
reflectance less than 10%, the procedure is the same except that the standards are sanded dry.
Further Reading
7
Chapter 6 Good Practice Guide 96 23

For readers wanting more information on colour measurement the following may be useful:-

Measuring colour, R W G Hunt, 3rd Edition, Fountain Press, Kingston-upon-Thames UK

(1998).
Billmeyer & Saltzmann’s Principles of Color Technology, R S Berns, John Wiley, New
York, 2000.
References
8
Chapter 6 Good Practice Guide 96 25

1 Taylor J, Hanson A R, Clarke P J, Campos Acosta J, Pons A, Czepluch W,

Jaaskelainen T, Hiltunen J, Gaudemer J, Soares O, Costa J L C,and Liedquist L, Good
Practice Guide to Surface Colour Measurements, European Commission, EUR19553,
Luxembourg (2000).

2 Commission Internationale de L’Eclairage, Colorimetry, CIE 15:2004, Vienna

(2004).

3 Judd D B and Wyszecki G, Colour in business, science and industry 2nd. Ed., Wiley,
New York (1982).

4 The expression of uncertainty and confidence in measurement, Edition 1, UKAS,

Feltham, UK (1997).

5 Malkin F, Larkin J A, Verrill J F and Wardman R H, The BCRA-NPL Ceramic

Colour Standards, Series II - Master spectral reflectance and thermochromism data, J.
Soc. Dyers Colorists 113, 84-94 (1978).

6 Verrill J F, Compton J A and Malkin F, Thermochromism of ceramic colour

standards, Appl. Opt. 25 (18), 3011 (1987).

7 Fairchild M D and Grum F, Thermochromism of ceramic reference tiles, Appl. Opt.

24, 3432 (1985).

8 Verrill J F, Knee P C and O’Halloran J, A study of the uniformity and

thermochromism of surface colour standards, CIE publication x011, 21-24 (1996).

9 Clarke P J, Hanson A R and Verrill J F, The determination of colorimetric

uncertainties in the spectrophotometric measurement of colour; Analytica Chimica
Acta 380 (2-3) 277-284 (1999).

10 Verrill J F, Clarke P J and O’Halloran J, Study of improved methods for absolute

colorimetry. Fifth report. Study of colorimetric errors on industrial instruments: NPL
Report COEM 2 (1997).

11 ASTM, Standard practice for computing the colors of objects by the CIE system,
ASTM E308-01 (2001).

12 Gardner J L, Uncertainty estimation in colour measurement, Color Res. Appl. 25,

349-355 (2000)

13 Ohno Y, A Numerical Method for Color Uncertainty, CIE Expert Symposium,

January 2001, CIE x020, Vienna, (2001)
Acknowledgements
9
Chapter 6 Good Practice Guide 96 27

This guide has had input from the partners of the European project “Harmonisation of
National Scales of Surface Colour Measurements” (SMT4-CT96-2140):
Dr. J. Taylor, Mr A R Hanson and Dr P. J. Clarke from National Physical Laboratory, NPL,
(UK).
Dr. J. Campos Acosta and Dr. A. Pons from Consejo Superior de Investigaciones Científicas,
CSIC, (ES).
Dr W Czepluch from Bundesanstalt für Materialforschung und Prüfung, BAM, (DE).
Prof. T Jaaskelainen and Mr J Hiltunen from University of Joensuu, JU, (FI).
Mr. J. Gaudemer from Laboratoire National d’Essais, LNE, (FR).
Prof. O Soares and Mr J. L. C. Costa from Centro de Ciencias e Technologias Opticas,
CETO, (PT).
Dr. L Liedquist Swedish National Testing Research Institute Ltd., SP, (SE)
Contact Details
10
Chapter 6 Good Practice Guide 96 29

Comments and feedback on the use of this guide should be sent to the author:

Dr Julie Taylor
National Physical Laboratory
Hampton Road
Teddington
Middlesex
United Kingdom
TW11 0LW

Telephone: +44 (0)20 8943 6539

E-Mail: [email protected]