This article was published in the February 2012 issue of LEDs Magazine.
View the Table of Contents and download the PDF file of the complete February 2012 issue.
Parts one and two of this series on LED photobiological safety focused on the potential hazards to the human body posed by exposure to optical radiation; the development of standards and their application; and the fine details of source evaluation.
This third part of the series focuses on the use of the IEC62471 standard to evaluate LED-based products. IEC62471 may be used to evaluate personal exposure to optical radiation, within, for example, the scope of the EU Artificial Optical Radiation Directive. However, its principal use is in providing a framework for evaluating the photobiological safety of finished products intended for sale on the market. The responsibility for ensuring that such an evaluation is performed resides with the manufacturer of the finished product, who, in many cases, is wont to reduce this burden.
It is clearly not possible to measure every LED in use, and indeed in many cases there is no need to do so. For example, the low visual response elicited from low-power white or colored LEDs leads one to reason that no photochemical safety concerns exist. However, as one considers LEDs of increasing optical power, the point at which one can no longer make such assumptions may not be obvious.
When are measurements required?
In the first instance, IEC62471 recommends that detailed measurements are not required for sources having a luminance less than 104cd/m2. This level is considered as one visually comfortable to view. The guidance is based on the expectation that, at this level, exposure limits will not be exceeded. However, it only applies to white or broadband sources emitting over the visible region.
Luminance does not fully take into account the emission of colored LEDs, nor does it take into account UV or IR emission. The luminance of a UV source may be below this level, yet one cannot use this information to base a conclusion on a potential UV hazard. In practice, this threshold luminance is particularly low, and is exceeded by many, even low-power, white LEDs.
Where the luminance of a white-light source exceeds this level, and for all other sources, one should proceed with the evaluation of photobiological safety, at the appropriate distance – 500 lx or 200 mm – depending on the intended application of the finished product.
General lighting service (GLS) sources are defined as white-light sources used to illuminate spaces. Within the context of LEDs, consideration is made of two technologies: phosphor-converted (PC) and color-mixed LEDs. Due to the narrow-band emission of LED chips, and the limited emission range of LED phosphors, one can restrict consideration to the visible region: no risks are posed in the UV or the IR.
Practically, the sole hazard in consideration is the blue-light retinal hazard, which dominates over the retinal thermal hazard for exposure times greater than ten seconds. It follows that it is the blue LED of both PC and color-mixed LEDs which gives the main cause for concern.
Consideration of the blue-light hazard of GLS sources is most conveniently demonstrated in evaluation of radiance through a measurement of irradiance, comparing the blue-light-weighted irradiance with the illuminance of the source (Fig. 1).
For a given illuminance, the higher the emission in the region of the blue-light hazard function, the greater the blue-light hazard posed. An increasingly prominent blue-emission peak lends a source a blue appearance, characterized by an increasing correlated color temperature (CCT). It can be demonstrated that at 500 lx, only LEDs having very high CCT (greater than approximately 10,000K) exceed the limits of the blue-light exempt risk group (RG), and that no sources will exceed blue-light RG1 (risk groups are discussed in Part 1). Since such high-color-temperature sources are seldom used in SSL applications, one can conclude that few GLS sources will pose any hazards at the 500-lx evaluation distance.
On the subject of GLS, two other points should be made. Firstly, with regard to certain sources, such as desk lamps and household spot lights (for which the determined 500 lx distance may be significantly greater than a likely exposure distance), the lack of clarity in the definition of GLS in IEC62471 has led to disagreement between laboratories of whether GLS or non-GLS measurement conditions should apply.
Secondly, and counter-intuitively, consider two GLS products, differing only in number of component LEDs or drive current. If the spectral output of both sources is the same, then the IEC62471 hazard evaluation is also the same, albeit performed at different 500-lx distances. Such a result may make sense where the two sources are used in distinct applications. However, in the not-uncommon case that they are marketed as alternatives for the same application, this further demonstrates that the evaluation at 500 lx is not a satisfactory point of reference.
The non-GLS category takes into account all types of LED, through the spectrum from the UV to the IR, including white LEDs used in non-GLS applications. Depending on the application, the optical output of such LEDs can vary significantly from very low-level indication to high-power LEDs used for example in industrial and signaling applications.
The analysis here is rather more detailed than is the case for GLS sources: at the close proximity of 200 mm, elevated risk-group classifications may indeed result, and there may be cause to consider multiple hazards for a single product. Table 1 provides an overview of the maximum reported RG of LED-based non-GLS products for each hazard considered by IEC62471. This excludes the thermal skin hazard, not part of the classification system.
In terms of the irradiance-based hazards, RG3 is certainly attainable, if not from a single LED, then by an array of LEDs. On the other hand, in the case of radiance-based hazards, since the measurement field-of-view (FOV) generally encompasses one, or a small number of, component LEDs, the maximum classification depends less on the collective effect of multiple LEDs in an array than on the output of individual LEDs.
While current blue-LED technology exceeds blue-light hazard RG1 by up to an order of magnitude, the RG2 limit is a further two orders of magnitude away. Furthermore, the often-cited fact that even the sun is an RG2 source would suggest that blue-light RG3 sources do not exist. Also, LED radiance is not sufficient to cause thermal damage to the retina; such damage can generally only be elicited by directly viewing certain lasers or arc lamps.
Analysis based on LED maker's data
In order to avoid the cost and effort of evaluating the photobiological safety of finished products, pressure has in the past been brought to bear on LED manufacturers to provide photobiological safety information which may be transferred to the finished product. It is clear that an IEC62471 evaluation of a bare LED is not directly transferable to a finished product, which may include multiple emitters and beam-shaping optics, so another strategy should be employed.
The irradiance of the finished product cannot in any way be predicted. However, in the case of radiance-based hazards, a measurement of the true radiance, coupled with the law of conservation of radiance, may be used to determine the maximum possible radiance of any finished product using a given LED.
IEC TR 62471-2 introduces this principle for the evaluation of the blue-light hazard (the dominant concern for retinal injury) through a measurement of true radiance of the component LED at 200-mm distance and 1.7-mrad FOV. The resulting value is adopted as the blue-light radiance of the final product, to be compared with the exposure limit values of each risk group in turn. It is important to note that care should be taken to ensure that the data provided by the manufacturer provides a correct analysis for the operating conditions of the finished product.
This procedure leads in many cases to an over-estimation of the hazard, since account is not taken of physiological radiance. This is demonstrated in Table 2, where a comparison is made between an IEC62471 analysis and a worst-case analysis of a particular product. In the former case, each RG is considered in turn, with measurements being performed in the correct FOV and compared with the RG exposure limit (resulting in an RG1 classification). In the latter case the worst-case radiance is assumed and compared with the limits of each risk group in turn (resulting in an RG2 classification).
A similar result is obtained in many instances, especially when considering high-power LEDs used in SSL applications. According to IEC TR 62471-2, blue-light RG2 requires the use of a warning label. This means that the lighting industry has been faced with the decision of either determining how to implement the recommendation of labeling, or not accepting such worst-case analysis evaluation, which clearly has no bearing on the true hazard posed by the source in the intended application. This procedure has generally been discontinued while awaiting a more acceptable solution, as will be seen below.
Analysis based on LED data-sheet values
Where no photobiological safety-evaluation information is available from an LED manufacturer, some have sought to make estimations based on data-sheet values, which typically report beam-emission angle and either total flux or intensity in photometric (lumen, candela) or radiometric (W, W/sr) quantities, depending on whether the LED emission wavelength is within or without the visible region.
Given the emission angle and the evaluation distance, the area illuminated by the LED may be determined and either total flux or intensity used to make an estimate of irradiance. To estimate physiological radiance, it is required to know both the intensity and the FOV area corresponding to the RG considered. Where intensity is not directly reported in the datasheet, it may be calculated from the total flux and beam-emission angle. In the case of white or colored LEDs, where photometric data is often provided, a conversion factor (lm/W) must be determined to convert to radiometric units.
In the case of hazards requiring the application of a hazard-weighting function, estimation without taking such into account represents an over-estimation. This errs on the correct side of caution, as befitting such an analysis. Again, care should be taken to ensure that the data provided by the manufacturer provides a correct analysis for the operating conditions of the finished product.
Should such a calculation indicate the existence of a classification that is higher than exempt, correct measurements are recommended. It need not be stated that the uncertainty associated with such estimations are necessarily high.
IEC TR 62471-2 also introduces the concept of mapping out the photobiological hazards associated with a source by determining hazard distance information to cover all potential applications. This procedure consists of the evaluation of a source at the minimum accessible distance, no less than 200 mm for the retinal hazards, and the determination (should any hazard be in excess of the exempt RG) of the distance from the source at which exposure is decreased to the required level for each remaining RG.
For irradiance-based hazards, this procedure is relatively straightforward, although one may be hampered by the requirement of measurement in a 1.4-radian FOV for all but the thermal skin hazard. The inverse-square irradiance law may be used with caution, but such calculations should not be necessary since irradiance can readily be measured at other distances by a number of techniques, such as the use of a luxmeter to seek an illuminance corresponding to the given level of irradiance sought.
Radiance-based hazards are more difficult to handle since measurements should be made in a specific FOV. Where the source subtends an angle less than the field of view, the hazard distance can be predicted since it will reduce with the square of the measurement distance, as the area of dark covered by the FOV increases.
Where a single emitter subtends an angle greater than the FOV, as a first approximation, physiological radiance will be constant until the distance at which the source subtends an angle less than the FOV. In the case of arrays, the physiological radiance may not decrease sufficiently before more LEDs fall into the FOV in which case, as a first approximation, physiological radiance will be constant until the distance at which the entire array subtends an angle less than the FOV (Fig. 2).
Hazard distance of LED luminaires
In awaiting an update of luminaire standards, evaluation of the photobiological safety of LED luminaires is currently performed through implementation of IEC62471. This situation has provided little satisfaction due to issues with the evaluation at 500-lx distance and the implementation of worst-case analysis to permit the transfer of LED manufacturer’s data. IEC committee SC34A is currently working on this issue, in considering the implementation of a restricted version of the hazard-distance analysis relative to the sole concern of white LEDs in GLS applications, namely the retinal blue-light hazard.
Based upon the assumption that light sources classified as exempt or RG1 for blue-light hazard are suitable for GLS applications, both component LEDs and finished products should, in the first instance, be evaluated at 200 mm in an 11-mrad FOV, with the spectral range extended to 300-780 nm to cover both blue light and photopic regions. This measurement serves as both an analysis of blue light RG1, and as a worst-case analysis, assuming that true radiance is measured.
Where the resulting blue-light radiance is below the RG1 exposure limit, the component LED or finished product may be considered exempt/RG1 in all conditions. Where the RG1 exposure limit is exceeded, the RG1 hazard distance should be determined through the evaluation of radiance as an irradiance measurement, with respect to a corresponding RG1 exposure limit expressed in blue-light-weighted irradiance. Given the ratio of luminance to blue-light radiance, the illuminance level at which the RG1 blue-light irradiance should be obtained is determined.
In the case of finished products, the distance at which this illuminance is obtained should be reported by using a luxmeter: at distances closer to the source than this distance, RG2 applies, elsewhere RG1/exempt applies. In the case of component LEDs, the illuminance value is simply reported in the data sheet such that the finished-product manufacturer can apply the aforementioned procedure to determine the RG1 distance for the particular product under consideration. This procedure is alas not quite as simple as it looks since the measurement should be performed in an 11-mrad FOV: not doing so will over-estimate the RG1 distance.
How luminaire standards will in future implement the RG1 hazard distance is still a work in progress, but it is clear that one can tolerate greater RG1 distances for ceiling-mounted applications compared for example to portable luminaires.
IECEE CB scheme & product marks
It is solely in Europe that IEC62471 has been implemented within a legal framework. However, IEC62471 has worldwide renown, through the implementation of the IECEE CB scheme and a wide range of product marks.
The IECEE CB scheme was set up to facilitate international trade in electrical equipment and is based on IEC product standards and a principle of mutual recognition of test results. Put simply, a manufacturer in country A, wishing to market his product in country B, need only have the product tested by a CB (certification body) testing laboratory in his home country. The CB test report will be accepted by the national CB (NCB) in country B and used to grant any required certification marks. Since 2009, testing to IEC62471 under the IECEE CB scheme LITE category (which requires testing to a number of other standards) is mandatory for LED-based GLS products.
In China, the mandatory CCC mark scheme requires testing of luminaires, for which IECEE CB test reports are currently accepted: through this certification route, testing to IEC62471 can be said to be mandatory for LED-based GLS products for sale in the Chinese market.
Furthermore, in many countries throughout the world, voluntary product-mark schemes are in existence. These are used to enhance the status of a product, by providing the consumer with increased confidence in its quality. Examples of such schemes include the UK BS kite, the German GS, the ENEC mark and the Korean KS mark, which are increasingly taking account of photobiological safety through application of IEC62471.
The revision of IEC62471/ CIE S009 by IEC TC-76 and CIE D6 is underway, yet given the proposed adjustment of certain exposure limits by ICNIRP (and the correction of the current retinal thermal hazard weighting spectrum) a publication date for the update has not yet been given. Two additional parts to the IEC62471 series are being drafted by IEC TC-76, including part 4, related to guidance on measurement methods. The CIE D6 has also formed working groups with terms of reference to consider the implementation of IEC62471.
In addition to this, the European Commission has asked the Scientific Committee on Emerging and Newly Identified Health Risks (SCENIHR) to look into the health effects of artificial light: it will be some time before this study reports its findings.
LEDs may not represent as great a hazard as lasers, yet given their widespread use and ever-increasing optical performance, it is correct that account should be taken of the potential hazards associated with these sources. One should also be aware that while IEC62471 is based on normal behavior, and the innate aversion response, consideration should be made of those overcoming the aversion response, particularly in the case of children who have a natural curiosity and no appreciation for potential hazards.
Reports of the potential effects of chronic blue-light exposure, resulting in age-related macular degeneration and loss of vision in the central visual field, are fairly well supported (but in need of much further research). However, there have only been a few contentious reports relating to acute exposure to LEDs, including ostensible retinal damage due to exposure to a violet LED and the suggestion that LEDs may be particularly dangerous for children, whose undeveloped lenses do not offer the retina sufficient protection in the UV.
As a last note to finished-product manufacturers, while little can be done to circumvent irradiance-based hazards (other than limiting access to the source), physiological radiance can be modified by design, by minimizing the optical power in a given FOV through appropriate spacing of the LEDs and using more low-power LEDs to do the job of a single high-power chip.