Aspects of Light Quality in Solid State Lighting by OSRAM Opto Semiconductors

Light-emitting diodes (LED) are considered to be the technology of the future for all applications in the field of general lighting because of their numerous, superior advantages: In addition to an extremely long lamp life of up to 100,000 hours, LEDs display particularly high energy efficiency and luminosity. One LED with just one chip shines with a luminous efficacy of over 120 lm/W. Furthermore, on account of their small size, these tiny light sources can provide ideal lighting solutions on a small area. However, the light quality of LEDs has been questioned, mostly in relation to three main aspects. First is the damage potential: Can the light from these small diodes damage artworks or other sensitive objects? The second aspect refers to the photobiological safety of LED systems, while the last focusses on the color rendering quality of LED light. These three aspects are examined in detail in the sections below.

LED Lighting: High Blue Component in the Light

LEDs are penetrating an increasing number of segments in general lighting. But alongside their frequently mentioned advantages, such as high efficiency and long lamp life, they undoubtedly are associated with problematic aspects, particularly in museum lighting, such as their ability to render color and their potential to damage objects on exhibit. The high blue component in the LED light spectrum and the potential risk are issues that repeatedly give cause for discussion. This article objectively explores and evaluates the damage potential of LED lighting as compared to conventional lighting solutions.

White light is produced in LEDs primarily by combining a blue LED chip with a phosphor converter. The blue light of the chip is partially converted by the right phosphor mixture into light with a longer wavelength and a yellow color. This light, comprising blue and yellow wavelengths, is perceived by the human eye as white light with the desired color temperature. Figure 1 shows the contribution of the blue chip in the light spectrum as a pronounced deflection at about 450 nm.

Figure 1: Typical spectra of a white LED and an incandescent lamp

Museum Lighting: Is LED Light a Risk?

From CIE 157:2004, a study of potential damage to museum objects by optical radiation, we know that short-wave optical radiation in particular displays an increasingly higher damage potential. The superficial comparison of the spectra of a white LED and an incandescent lamp in figure 1 gives the impression that the white LED has a significantly higher blue component, making it appear more harmful than the incandescent lamp. However, this is only apparently so. To make a well-founded statement about the damage potential, the two spectra must be analyzed and compared on the basis of CIE 157:2004.

CIE 157:2004 describes a suitable method for evaluating and estimating the damage potential of optical radiation to museum objects. The core of the evaluation is the weighting of the spectrum of the light in terms of the relative spectral object sensitivity.

Together with the experimentally determined threshold radiation values, it is possible to estimate the damage potential of different spectra and light sources. The damage potential is the ratio of object-damaging irradiance to illuminance. As an example, object-damaging irradiation is calculated for the relative spectral object sensitivity of watercolor on paper (Figure 2). In the first step, several white LEDs with different color temperatures and phosphor mixtures are compared with the incandescent lamp mentioned above. As shown in the comparison with an incandescent lamp in figure 3, the damage potential of the white LED is even lower, depending on the color temperature, despite its high blue component. Consequently, illuminating museum objects with LED light is less damaging than with an unfiltered incandescent lamp.

Figure 2: Relative spectral object sensitivities for different materials

Figure 3: Comparison of damage potential of different light sources

LED: The Ideal Light Source for Museums

In museums, light sources often are equipped with special UV and infrared filters to minimize damage to the objects. To be able to classify the damage potential of LEDs in valid fashion, a representative lighting situation in a museum was evaluated. In the selected museum, the basic standard lighting consists of wall-washers and accent lighting provided by spotlights. The light source for the basic lighting is a fluorescent lamp; a halogen lamp is used for the spotlighting. Both are fitted with the standard museum UV and IR filters. For the study, this lighting was replaced by LED lighting. The spectra of both lighting solutions were scaled to an illuminance of 50 lux, and their damage potential on a watercolor painting on paper evaluated.

A comparison of the two lighting solutions in figure 4 shows that the damage potential of the LED lighting is similar to, or even lower than, that of the existing, high-quality museum lighting solution based on fluorescent and halogen lamps, given comparable color temperature. Daylight has the highest damage potential, which is why many museums attempt to ban it entirely from their exhibition spaces, so as to minimize as much as possible any impairment of the objects on display.

Figure 4: Comparison of the damage potential of different lamps

The study shows that modern light-emitting diodes are excellently suited as light sources for high-quality museum lighting. Although the obviously high blue component in the spectrum of white LEDs is deterring at first, closer consideration reveals that they pose no threat to the sensitive objects. Apart from the low damage potential, the high luminous efficacy combined with the long lamp life makes LEDs the perfect light source for the museum lighting segment.

IEC 62471 on Photobiological Safety

After examining the impact of LED radiation on objects, we turn now to its effect on the human eye. Lamp safety standard IEC 62471 defines safety guidelines for lamps and lamp systems, referred to as photobiological safety. The aim of this standard, accepted by lamp manufacturers, is to enable a standardized estimation of the magnitude of the risk to the human eye of potential radiation from lamps and lamp systems. IEC 62471 defines exposure limits, reference measurement categories and the classification scheme for the evaluation and control of photobiological hazards in the wavelength range from 200 to 3,000 nm. The defined numbers are applicable to most lamps and lamp systems, including luminaires: LEDs as well as incandescent, fluorescent, high-pressure discharge and other lamps can be evaluated, but not lasers. According to the IEC 62471 guidelines, both LED lamps and conventional light sources are classified in the same risk group, as shown in figure 5. Traditional lamps are in the upper part of the diagram, various LED lamps along the bottom. This shows that LEDs are just as safe as other conventional light sources and do not damage the human eye. In other words, nothing stands in the way of their use in the general lighting segment.

Figure 5: Classification of conventional and LED lamps in risk groups to IEC 62471

Color Rendering: A Challenge for Light Technologies

The greatest challenge for LEDs and other light technologies, however, is the adequate rendering of colors, as the following example illustrates. Every shopper has at some time faced the dilemma of wanting to buy a piece of clothing, but also wanting to first see it in daylight. The reason: Under the light in a store, colors often look different than they do at home or outdoors in daylight (Figure 6). The following section explains why color rendering differs so much, and which advantages and challenges exist, particularly for LEDs.

Figure 6: The color of products in a store often looks entirely different at home

We are surrounded by any number of artificial light sources every day. We use primarily incandescent and halogen lamps at home, fluorescent lamps at work. Stores use spotlights fitted with high-pressure discharge lamps to illuminate their latest products, and xenon lamps in car headlights light up the roadway for us at night. But the LED also has established itself in recent years as an important light source in these fields. But all of these artificial light sources have different spectral power distributions. If a colored object is illuminated by different light sources, the color rendering can differ, and the object looks slightly different depending on the lighting.

If a colored object is illuminated with a specific light, for example an incandescent lamp, the illuminated object reflects some of the spectrum. The eye and brain of the observer then evaluate the reflected spectrum, which ultimately results in a specific color impression. If the same object is now illuminated with light of yet another spectral makeup, for example a fluorescent lamp, the colors of the object may look different again. This is one of the reasons why colored objects, like clothing, look different under different light sources. To determine and predict how a colored object will appear based on the different spectral power distributions, the Commission Internationale de l’Éclairage (CIE) created the Color Rendering Index.

CRI – Standardization Method for the Color Rendering of Light Sources

As early as in 1974, CIE published the first detailed method that made it possible to express the rendering of colors under different lighting conditions as a number. For this purpose, CIE defined eight standard and six special samples as test colors. In this connection, the calculation of the CRI always involves a comparison of a test light source with a reference light source. If the correlated color temperature of the test light source is under 5,000 kelvin, the reference light source is a Planckian radiator with the same correlated color temperature. At a color temperature of over 5,000 K, a spectrum close to daylight is selected as a reference.

The test colors are then illuminated virtually by the test light source and the reference. The resulting tristimulus values X Y Z are derived and converted to the U*V*W* color space. Chromatic adaptation is carried out using a von Kries-type chromatic adaptation transform. The chromacity difference for each test color in the U*V*W* color space can then be calculated. The 8+6 color rendering indices Ri are calculated based on the chromacity difference Ri = 100 – 4.6ΔEi. The general color rendering index is the average of the first eight individual indices: Ra = 1/8 Σ Ri.

Figure 7: Example calculation of a Ra value of 62 using the color rendering index (CRI)

CRI 100 – The Optimum in Color Rendition

In the example calculation in figure 7, the color rendering index is calculated for a white LED with a high correlated color temperature of about 5,700 K. Because the color temperature exceeds the value of 5,000 K, daylight is used as a reference light source. The test color samples are illuminated by both the white LED and the reference light source (daylight), whose spectra differ significantly, as the top section of the figure shows.

The different color coordinates resulting from the lighting are represented in the Lab color space diagram, where the test lighting is shown by the blue line, the reference by the gray. In the diagram, the blue and gray points and lines differ significantly, because the resulting color coordinates of the test colors differ. The rule in this case: The greater the distance in between, the lower the Rx values. Therefore, the red R9 color field can also be negative on account of the general scaling of the color rendering index. In the visual impression of color difference based on the stack of colored towels, the difference in color rendering is clearly noticeable at a Ra value of 62 for the test and reference light sources.

To achieve very good color rendering close to the optimum value of 100, the color coordinates of the test and reference light sources must be as similar as possible or even identical. In this case, color perception is not distorted and color rendering accordingly good. If we now compare a warm white LED with 2,600 K, optimized for high CRI, with a Planckian radiator of the same CRI, the test colors look very similar (Figure 8).

Figure 8: Example calculation of a Ra value of 97 using the color rendering index (CRI)

The points and lines in the Lab color space then match more or less, which indicates Rx values that lie close to optimum color rendering, e.g. in this case 97.

Establishing a Precedent: Color Fidelity Versus Color Preference

However, a high color rendering value can be achieved even with a very different spectral power distribution between the test and reference spectra, so that colors look very similar. In what is known as the Brilliant Mix Concept, the light from a greenish-white and a red LED is mixed. The color points in the Lab color space resulting from the test and reference light sources are very close together. With a small deviation, a value of 91 can be calculated for the color rendering index. However, the color coordinates of the Brilliant Mix lighting are always on or outside the reference light source in this case. As a result, the colors are slightly changed and appear more brilliant, saturated and glossy, although the color rendering index is not at 100. This effect is also referred to as color preference: It describes how strongly the saturation of a color is increased by the spectrum of the lighting. Within certain limits, relatively high saturation is even preferred. However, color preference has not yet been incorporated in the existing color rendering index definition.

These considerations show that the color rendering index in its current form has reached its limitations in some areas. For this reason, work on revising it is already underway: CIE is discussing what possible routes can be taken to calculate and define the color rendering index. It must be accepted, however, that compromises will always have to be made between color fidelity, color preference, custom spectral power distribution and efficiency. Furthermore, some fields of application, such as street lighting, do not require a high color rendering value. Fortunately, the spectrum of LED light can be modified in many directions, meaning that the optimum LED can be selected for every lighting requirement.

Conclusions

Closer examination shows that in addition to possessing the advantages described in the opening paragraphs, current LEDs are in no way inferior to conventional light sources when it comes to light quality and color rendering. What is more, research and development in the field of LED technology is extremely fast-paced and efficient, leading to the continuous optimization of these miniature light sources. In view of these results, LEDs are justifiably referred to as the light technology of the future.