Color Measurement of Fluorescent Colors – a CHALLENGE

1. Introduction

Consistent color and appearance stand for quality. Particularly in the case of products consisting of several parts made of different materials and produced at different locations, a uniform color impression can only be guaranteed if all individual parts are checked for color quality using a spectrophotometer.
As soon as fluorescent ingredients are used, the otherwise unique reflectance curve (spectral fingerprint) is dependent on the light source used in the spectrophotometer. Therefore, the measurement results from different spectrophotometers are no longer comparable and the correlation with the visual impression under a certain type of light is only guaranteed to a limited extent. Following, it will be explained why standard spectrophotometers are only suitable to a very limited extent for assessing the color quality of materials with fluorescent ingredients, which problems are associated with this and which measurement technology now available on the market will provide a remedy in the future.

2. Description of Fluorescence

2.1    Theory of Fluorescence 
The name fluorescence is derived from the fluorescent mineral fluorite (fluorspar, calcium fluoride, CaF2). Fluorescence is the property of atoms and molecules, so-called fluorophores, to absorb light at a certain wavelength and then emit light at a different wavelength. The difference between the excitation wavelength and the emission wavelength is known as the Stokes shift. The emitted fluorescent light is usually shifted into the long-wave range of the light spectrum compared to the excitation light. 
What is this effect based on? The principle is explained in Figure 1. Most molecules assume the lowest energy state at room temperature, the so-called ground state (S0). Within this ground state there are different vibrational levels. Before they are excited, many molecules adopt the lowest vibrational level. When a molecule absorbs a certain wavelength of light (blue vertical arrow), the absorbed photon causes the molecule to assume a higher vibrational energy state (S1 and S2). The molecules then collide with other molecules, causing them to lose their vibrational energy (red curved arrows) and return to the lowest vibrational level of the excited state. The molecule can then return to the ground state (S0)[1].
When the molecule returns to the ground state, it emits a photon of light with a different wavelength than the one with which it was excited. This is the moment when the molecule shows fluorescence (green vertical arrow). Chemists refer to molecules that can show fluorescence as fluorophores. [2] The states and transitions of electrons in a fluorophore are complex but can be visualized with the help of the Jablonski diagram (Fig. 1).

2.2    Types and Industrial Applications of Fluorophores

Fluorescent molecules and materials come in all shapes and sizes. Some are naturally fluorescent, such as chlorophyll. Others are molecules that have been specially synthesized as stable organic dyes that are added to otherwise non-fluorescent systems. Depending on their size and structure, organic dyes can emit from the UV to the near IR. In general, commercial fluorescent dyes can be divided into three main groups: inorganic fluorophores, optical brighteners and daylight fluorescents.

Inorganic fluorophores: Even though the term “fluorescence” is based on the mineral fluorite (CaF2), it was later shown that the observed fluorescence did not originate from the fluorite itself, but was caused by impurities with divalent europium. Europium belongs to the group of lanthanides and is therefore a rare earth metal. The fluorescence of europium has a very large Stokes shift and is also extremely photostable. [3] The element is frequently used in security-relevant products, such as banknotes and ID documents, and also gives the euro banknote the desired counterfeit protection (Fig. 2).

Optical brighteners: The best-known group of fluorophores are the so-called optical brighteners, which are most frequently used in the textile, paper and plastics industries. Optical brighteners are substances whose excitation maxima are in the UV range (usually between 300 - 400 nm) and emit in the range of 400 - 480 nm, ideally between 430 - 450 nm. This means that non-visible, incoming UV radiation from daylight is re-emitted in the short-wave, blue range of visible light. The emission of additional blue light compensates for a slight yellow hue of the base material and thus increases the degree of whiteness. In some cases, the reflectance exceeds 100%, making the base material appear “whiter than white”. The optical brighteners currently used can be divided into six groups. The so-called stilbene compounds form the largest group, accounting for around 80% of all optical brighteners produced. [4]

Daylight fluorescent dyes: Daylight fluorescent dyes have both excitation and emission maxima in the visible range of the spectrum, meaning that almost any light source can excite the materials. They are used in many areas, from light-colored papers and paints to plastic products. They are very effective because they increase the light yield at certain wavelengths, in some cases far beyond what can be generated by reflection alone. A red, yellow or orange material with an apparent reflectance of over 100 % R can be produced by a combination of reflection, the non-fluorescent component and emission of the fluorophore (Fig. 3).

2.3    UV-Resistance and Aging Behavior  

Materials and paints with fluorescent ingredients are in general considered to be less resistant to light. The lightfastness of paints and coatings is influenced, for example, by the pigments, the binder and the thickness of the paint application. The more pigmented and the thicker the coating, the higher the lightfastness. However, there are limits to both factors, as the fluorescence effect is limited if the pigmentation is too high, and the layer thickness is too thick. 
Daylight fluorescent inks in particular are sensitive to light, UV radiation and heat. UV-induced degradation of the daylight fluorescent ink leads to a loss of fluorescence and fading of the chromophores. The loss of fluorescence is noticeable in white by yellowing and in other shades by darkening. The destruction of the chromophores, the chemical structures that give color, leads to fading of the pigments. [5]

Image 1 Jablonski-Diagram (top) and spectral data as Gaussian normal distribution with focus on excitation & emission maxima (bottom)

Image 2 Deutsche Bundesbank, Central Communications Division, “The new 20 Euro banknote”, February 2015

Image 3 Spectral curve yellow and luminous yellow

3. Measurement Principle of a Traditional Spectrophotometer

There have been and still are different types of spectrophotometers, but they all have one thing in common: the main purpose of a spectrophotometer is to characterize a material with respect to its spectral properties. This is achieved by applying light to a sample and then measuring the intensity of the light after it has interacted with the sample.

When measuring a fluorescent sample with a spectrophotometer, the light source used to illuminate the material plays a fundamental role. At the current state of the art, three different types of lamps are used for illumination: Tungsten halogen lamps, xenon flash lamps or LEDs. If a spectrophotometer uses a polychromatic light source, the spectrum of the light must be spread out, i.e. evenly divided according to wavelength, which is done by a monochromator, usually a diffraction grating. A diffraction grating is a glass or metal plate with very narrow parallel lines that generate a spectrum through diffraction and interference of the light. The fanned-out spectrum is then measured wavelength by wavelength using sensors such as photodetectors. Photodetectors measure light by converting incoming photons into an electrical signal. The result is the respective radiation intensity or remission for each measured wavelength range (Fig. 4).

The sample surface is usually illuminated in the visible wavelength range of 400 - 700 nm. High-quality spectrophotometers whose light sources have a UV component can be equipped with a so-called UV cut-off filter in order to limit the energy of the illumination to the visible spectrum. Spectrophotometers whose light source has no UV component are sometimes supplemented by an additional UV light source (e.g. a UV LED). By comparing two measurements, one with and one without UV component, it should be determined whether a material contains fluorophores. This comparison only allows quantitative conclusions to be drawn about fluorophores that are excited in the UV range.

Image 4 Functional principle of a spectrophotometer

4. Limitations of a Traditional Spectrophotometer

4.1    Lack of Differentiation between Remission and Emission
The remission of an object determines its color. In a non-fluorescent molecule, photons are absorbed and immediately return to their original energy state. The energy is released as heat or simply a photon of the same wavelength is emitted (Fig. 5). In simple terms, the light of a specific wavelength that strikes the surface is reemitted at the same specific wavelength. Spectrophotometers are used to analyze precisely this process. The result is a typical remission spectrum, also known as a spectral curve, which is unique to the object. However, if fluorophores are also included in a material, a traditional spectrophotometer cannot differentiate between the remission of the non-fluorescent materials and the emission of the fluorescent components. The result is therefore a combination of reflectance and emission spectrum, which can lead to quality control problems. [6]

4.2    Dependency on Built-in Illumination
Figure 6 shows an example of the spectral power distribution of a xenon flash lamp and a tungsten halogen lamp (Fig. 6). While the tungsten halogen lamp emits almost no photons in the non-visible UV range, the xenon flash lamp has a lot of energy in this range. The two light sources also differ in the short-wave, blue range of the visible spectrum: The xenon flash lamp emits considerably more energy onto the sample surface than the tungsten halogen lamp. Depending on the specific characteristics of the fluorophores contained in the material, this results in different levels of excitation and therefore different levels of emission. From this we can deduce that the reflectance spectra of the same fluorescent sample detected with different spectrophoto-meters clearly differ depending on the light source used. A traditional spectrophotometer is therefore not able to evaluate the combination of reflectance and emission spectra of a fluorescent material independently of the illumination used in the instrument.

Initially, an example using the spectral curves of optical brightened paper, detected by two traditional spectrophotometers with different light sources. It can clearly be seen that the UV component of the xenon flash lamp excites the optical brighteners much stronger than the tungsten halogen lamp and therefore emits significantly more energy in the blue wavelength range. The maximum reflectance measured with the tungsten halogen lamp is around 95% (orange line), the measured value of the xenon flash lamp (gray line) is around 120% combined reflection and emission (Fig. 7). Consequently, the CIELAB measurement results of fluorescent samples obtained with different spectrophotometers cannot be compared. 

The following table uses CIELAB measurement results on a pair of slightly fluorescent coil coating samples to show that even a small amount of fluorescence can lead to considerable differences. The result obtained with the xenon flash lamp spectrophotometer, for example, could lead to a “pass” in quality control, while the instrument with the tungsten halogen lamp lead to a “fail” for the sample pair (Table 1).

Image 5 Absorption/Remission of Material without fluorescence

Image 6 Spectral Power Distribution Xenon-Flash and Tungsten-Halogen Lamp

Image 7 Brightened paper measured with spectrophotometers from different manufacturers

Image 8 Comparison of different spectrophotometers on a fluorescent sample pair

5. Methods for Examining a Material for Fluorescence

5.1    Measurement With and Without UV Component
Due to a lack of suitable measurement technology for the industrial quality control of fluorescent sample materials, the state of the art to date has been to carry out two series of measurements for a pair of samples - one with and one without UV component. If the CIELAB values obtained in the two series of measurements are different, it can be assumed that the material contains fluorophores that are excited in the UV range.  The disadvantage of this method is obvious: on the one hand, it is more of a relative statement than an actual measured variable. Secondly, only a statement can be made about fluorophores whose excitation occurs in the UV range. The wide range of fluorophores that are excited in the visual range are completely disregarded in this type of analysis.

5.2    Fluorescence Spectroscopy (Fluorimeter)
In science, fluorescence spectrometers are used to excite fluorophore molecules and measure their emitted fluorescence. Fluorescence spectroscopy analyzes the fluorescence of a molecule based on its fluorescent properties. Light is directed into the fluorescence spectrometer using an illumination source, preferably a xenon lamp with the broadest possible spectrum down to the deep UV range (~ 200 nm). The light passes through a monochromator, which filters out a specific wavelength, and then the light is directed onto the sample to be analyzed. The sample emits an emission spectrum which is recorded by a downstream spectrometer. To measure fluorescent samples, this process is repeated one after the other from 360 nm - 700 nm in one nanometer steps. Depending on the desired signal quality and resolution, a single measurement takes approx. 1 second so that the evaluation of the entire wavelength range 360 - 700 nm can take up to 2 hours. The result is 340 emission spectra (Fig. 8).

A graphical evaluation that is frequently used is the so-called excitation-emission matrix (EEM). An EEM is a 3D scan that produces a contour plot of excitation wavelength, emission wavelength and fluorescence intensity. EEMs are used for a variety of applications and are often referred to as a “molecular fingerprint” for many different types of fluorophore samples. [7] In the broadest sense, they could be compared to the spectral fingerprint of a non-fluorescent sample captured by a spectrophotometer. Fluorescence spectroscopy has a wide range of applications in research and science. For example, the method is used in biomedicine for the investigation of proteins, nucleic acids and living cells. Fluorescence spectroscopy in industry, especially for quality control, is used almost exclusively in the pharmaceutical and food technology sectors. As the acquisition costs for a fluorescence spectroscope (fluorimeter) are considerably higher than for a spectrophotometer and the measuring time is also exponentially longer.

Image 9 Donaldson / EEM Matrix

6. Spectrophotometer & World's Smallest Fluorimeter in One

As already described in the previous chapters, measurement results obtained with different spectrophotometers on fluorescent material are only comparable to a limited extent or not at all, depending on the physical design and built-in light source of the measuring instrument and the specific excitation characteristics of the fluorophores. However, quality control of fluorescent materials on an industrial scale with a fluorimeter is hardly possible due to the long analysis time and high costs. So how can the industrial quality control of fluorescent material be realized?

6.1    Requirements for Instrumental Measurement of Color and Fluorescence
BYK-Gardner has set itself the goal of enabling industrial quality control of fluorescent materials such as plastics, paints, paper and textiles with a completely new and innovative method that matches the visual impression under all light conditions. For this purpose, a spectrophotometer was to be combined with a miniaturized fluorimeter in a handy - if possible, even portable - measuring instrument. The measurement time for each individual measurement should be only slightly longer than is the case for the measurement of a non-fluorescent solid color. The purchase costs should also be within the range of a traditional spectrophotometer for solid colors.

This target was realized in two instrument families: spectro2guide is a portable spectrophotometer that is available with a 45°c:0° and a d:8° (spin/spex) measuring geometry. The color2view is a table-top spectrophotometer with 45°c:0° measuring geometry. In the following text, both device names are omitted to simplify readability.

6.2    Physical Design and Measurement Principle
It has been standard practice at BYK-Gardner for 30 years, to use LEDs as light source in all instruments, as they have excellent short-term, long-term and temperature stability and thus provide the basis for outstanding repeatability and instrument agreement. To provide polychromatic light for the spectral color measurement, the two new instrument families use so-called “Full Spectrum”-LEDs with three phosphors, which have been specially developed to imitate the spectrum of natural sunlight as precisely as possible (Fig. 9). [8] 

To evaluate the fluorescence, the measuring principle of a fluorimeter was adopted in a simplified form. 12 monochromatic LEDs each equipped with a narrowband interfernce filter are used as the photon source in the portable measuring device  (360 - 660 nm) while the benchtop spectrophotometer uses 18 (300 - 700 nm), which illuminate the sample surface sequentially (Fig. 10). For each individual measurement with the monochromatic LEDs, the spectral power distribution of the excitation is calculated and set in relation to the mesured spectral power distribution of the emission.

In contrast to a traditional spectrophotometer, not just one but up to 19 spectral curves are detected on a sample containing fluorophores: one illuminated with polychromatic light and 18 others illuminated with monochromatic light. The two graphs below show as an example the excitation spectrum of the first five monochromatic LEDs (Fig. 11) as well as the remission at the same wavelength and the emission in the lower-energy, long-wavelength range (Fig. 12) for the sample RAL 1026 “Luminous yellow”. The excitation maximum of LED 02 (yellow), for example, is at 415 nm and therefore in the visible range. However, only a fraction of the original excitation energy is reemitted at the same wavelength, the majority of the energy is emitted between 490 - 650 nm with a peak at 530 nm.

By combining a spectrophotometer and fluorimeter, the two instrument families spectro2guide and color2view are able to differentiate between remission and emission. This is the basis for the internal calculation of the so-called “zero curve”. The zero curve corresponds to the spectrum of the measured material without a shift from excitation to emission wavelength and thus enables the evaluation of a fluorescent material independently of the spectral power distribution of the light source installed in the instrument. For the first time, the “Zero-Curve” enables the calculation of correct L*a*b* values depending on the selected standard illuminant for fluorescent sample material. For this purpose, the known excitations are set into relation to the selected illuminant and the remission spectrum is corrected using the correspondingly weighted emission spectra for each monochromatic LED. Thus, guaranteeing an excellent match with the visual impression.

Image 10 Full Spectrum LED with three phosphors used for BYK devices (left), Commercial “Full Spectrum”-LED with only two phosphors (right)

Image 11 Monochromatic LEDs in the portable spectro2guilde (left) and benchtop color2view (right)

Image 12 Example of the calculated excitation spectrum of 5 monochromatic LEDs of the spectro2guide

Image 13 Emission spectrum of 5 monochromatic LEDs measured for sample RAL 1026

Image 14 Luminous Yellow measured with spectro2guide, Standard Illuminants A and D65 and zero curve

7. Fluorescence Data Analysis or Prediction of Longterm Color Stability

7.1    Fluorescence Slider
The software included with both families of measuring instruments enables a detailed graphical analysis of the specific characteristics of the fluorophores contained in the material. The desired monochromatic LED (excitation wavelength) can be selected using a slider and the corresponding emission spectra are displayed immediately. If you compare the sample (right) with the standard (left) at the excitation wavelength of 570 nm in the graph below, you can see that the emission spectra differ significantly. While the emission spectrum of the fluorophore in the standard shows two clear peaks at 620 nm and 670 nm, the fluorophore in the sample hardly emits any light over the entire range from 590 - 760 nm. It can therefore be clearly demonstrated that the material composition of the standard and the sample differ with regard to the fluorophores used and thus with respect to the optical impression and possibly also in terms to their long-term stability.

The analysis options using the slider can be summarized as follows: 

• A material that contains no fluorophores shows a zero line because no light is emitted. 
• Materials that contain the same fluorophores show the same emission spectra, as the emission properties are identical. However, the emission spectra may differ in intensity depending on the amount of fluorophores added. 
• If the shape of the emission spectra differ from each other, it can be assumed that different fluorophores were used. 

7.2    Special Indices dE Fl and dE zero
For simplified evaluation, two new indices were developed to quantify the fluorescence energy: dE FL and dE zero.

dE Fl describes the change in the visual color impression of a material, assuming that all fluorophores contained have decayed. It is therefore not a classic “sample to standard” comparison, but a comparison of a sample in its “actual” state to a predicted state in the future. For example, the optically brightened, white sheet of paper now compared to the yellowed sheet of paper as soon as it was exposed to the sun.
dE zero describes the color harmony between a standard and a sample under the assumption that all fluorophores contained in both have decayed. Imagine several white plastic windows on the front of a house: For the customer, it is not only important that the windows show the same whiteness at the time of installation, but also that they change uniformly over time. If one window yellows more than all the others because fewer or different fluorophores were used, the future color harmony decreases. The example for a white coil coating in Fig. 15 shows that dE zero (future color harmony) is significantly higher than dE (current color harmony). 

The calculations of the dE Fl and dE zero indices are both based on the color difference formula used by the customer. If, for example, ?ECMC or ?E00 is used for the quality control of the solid colors, the result for dE Fl and dE zero is based on the same equation. The result of both indices therefore corresponds to the visual color impression: the higher the value, the greater the predicted color difference after the degradation of all fluorophores. As for the standard color differences, a limit can also be set for dE Fl and dE zero in the standard management.  
Both new indices are not only shown in the software, but also on the instrument display. As an additional visual indication, a status LED on the instrument blinks blue if fluorescence has been detected and pink if the defined limit for fluorescence has been exceeded.

Image 15 Fluorescence Slider in smart-chart Software

Image 16 dE Fl and dE zero in smart-chart Software

8. Comparison of Results to Reference Fluorimeter

Following, a comparison is made between the results of a laboratory fluorimeter from Horiba and those of the color2view spectrophotometer for fluorescence. An exemplary sample pair (standard/sample) made of white-coated metal (coil coating) is measured for the comparison. This was selected because, on the one hand, different fluorophores were used in the standard and sample and, on the other hand, the fluorophore used in the standard has very specific characteristics. To ensure better comparability, the results of both instruments are displayed as an Excitation-Emission Matrix (EEM).  
The fluorophore included in the standard shows very steep slopes and five individual emission peaks at the wavelengths 620, 640 and 670 nm in the graphical evaluation of the Horiba results (Fig. 16). Steep slopes as well as multiple peaks pose a particular challenge for any measuring system, as the prerequisite for their detection requires a very fine scanning grid for excitation and emission. A comparison of the graph resulting from the color2view measurement result (Fig. 17) shows a very high level of agreement. The miniaturized fluorimeter in the color2view is also able to localize the 5 individual emission peaks at the corresponding wavelengths.

The most obvious difference between the two evaluations is the height of the emission maxima at 640 nm. While in the evaluation of the Horiba fluorimeter one of the five peaks is clearly more distinct than the other four, the emission maximum determined with the color2view is at a similar height to the two flanking peaks. The reason for this again lies in the very steep slope of the peak with the pointed end. To be able to display the final emission maximum, the detection must take place exactly at the required wavelength. As the laboratory fluorimeter from Horiba detects with a tighter grid than the miniaturized fluorimeter in color2view, the wavelength of the emission maximum is hit precisely.
This deviation in the characteristics of the emission maximum is negligible with regard to the calculation of the dEFl and dEzero indices, as both indices are calculated based on the integral, i.e. the area under the curve. Accordingly, the smart-lab data analysis software does not display the emission as a 3D model (EEM) but as an emission curve depending on the selected excitation wavelength (Fig. 18). Details on the fluorescence slider can be found in chapter 7.1.

If you compare the standard with the sample in Figure 16 (Horiba) or 17 (color2view), you can see in both figures that the standard emits significantly more light than the sample. From this, it can be concluded that the standard changes more in visual terms than the sample after the fluorescence has decayed. This assumption can be confirmed by the newly developed fluorescence indices. The dEzero calculated for the standard with a value of 0.67 is almost twice as high as that of the sample with a dEzero of 0.3. If one compares the current color harmony of standard and sample (dE94 = 0.83) with the prediction for the future harmony after all fluorescence has decayed (dE94 zero = 1.66), a clear deterioration will also occur here.

Image 17 MME Horiba Laboratory Fluorimeter (Standard left, Sample right)

Image 18 MME color2view Benchtop Spectrophotometer (Standard left, Sample right)

Image 19 Fluorescence Slider, smart-lab Software, Excitation 570 nm, Emission 590 – 760 nm

Image 20 dE Fl and dE zero in smart-chart Software

9. Summary

The new generation of instruments, spectro2guide (left) and color2view (right), combines a conventional spectrophotometer with a fluorimeter, thereby revolutionizing quality control (Fig. 18): On the one hand, the instruments can clarify with a single measurement whether fluorophores have been used in a material - perhaps without knowing it - and thus explain deviations between different classic spectrophotometers. On the other hand, the new generation offers for the first time an affordable and at the same time accurate fluorimeter on an industrial scale, which enables the quality control of fluorescent material independently of the light source installed in the instrument and thus provides comparable CIELAB measured values. Finally, the new dE Fl and dE zero indices provide a prediction of how much a fluorescent sample is likely to change due to exposure to sunlight or how the future color harmony between two samples will change after decay of all fluorescent components.

Image 21 Portable spectro2guide (left) and Benchtop instrument color2view (right)