Novel compact LED multiplexing methods can reduce the space required to meet illumination requirements.
Ben Standish, Bookham Inc.
The much-anticipated availability of LEDs for applications in biophotonics finally is manifesting itself in a new selection of more efficient, longer-lived illumination products. LEDs have long held promise as a simple, low-cost alternative to lasers to improve upon the shortcomings of bulb-based illumination — including short lifetimes, excess heat production and the associated energy wasted in generating it as well as, often, an emission spectrum that is too broad.
Successful implementation of LEDs into biomedical instruments is not without serious challenges because Lambertian emissions are inherent in LEDs and because the wavelengths available that are bright enough for target applications are limited. Traditional methods of collecting LED flux and of directing it in a useful way usually involve large and often expensive optics and/or poor efficiency. However, innovation directing LED output is providing novel alternatives (Figure 1).
Figure 1. This scale drawing contrasts the size of compact, enclosed LED combining next to a larger conventional free-space approach. Both methods combine different colors of light with dichroic filters, but the smaller method contains the LED flux in geometry-optimized channels of high-reflector coated glass.
Lambert’s cosine law states that the radiant flux emitted is directly proportional to the cosine of the angle between the measured angle and normal incidence (Figure 2). In practical terms, this means that too much of the light is emitted to the sides rather than straight ahead from the LED emitter surface. This is fine for ambient room lighting but problematic when imaging light upon a relatively small area such as the fiber bundles, liquid light-guides or microplates used in biomedical instruments.
Figure 2. LED radiometric flux is challenging to capture because of its Lambertian emission pattern. Although LED packagers attempt to mitigate this, generally speaking, LEDs have the same radiance when viewed at angles. The number of photons directed into any wedge is proportional to the area of the wedge.
These challenges have been addressed in consumer electronics video projection optics that have a small digital micromirror or liquid crystal device serving as the engine driving many high-definition rear- and front-view projectors. Unfortunately, the specific requirements for illumination in biophotonics rarely allow the same video display optics to cross over to the domain of life sciences.
White-light applications for visual illumination and narrow-spectrum illumination for fluorescence excitation are two areas in the life sciences where LEDs hold much promise. White-light and fluorescence excitation have nearly opposite spectral requirements, but LEDs can meet these by multiplexing colors into one path. For white-light applications, three colors are on at the same time. For fluorescence, only one LED is on at a time, but all wavelengths still must be projected into the same optics.
Achieving a bright white
White-light applications are simpler from an optical design perspective. Bulb-based illumination long has been the only solution, but it is not without drawbacks. Bulbs for these applications typically last only 300 to 2000 hours, depending on the type, whereas the instruments themselves have useful lifetimes of tens of thousands of hours. In-field replacement of a bulb, which often requires recalibration of instruments, is a costly nuisance. On the other hand, LEDs can achieve 20,000+ hours when adequately driven and cooled; they also are more efficient, resulting in less heat and in fewer related thermal management issues. Their instant-on nature is a welcome alternative to bulb warm-up times.
Not all is wonderful with white LEDs, however. Not only is overall brightness well below that of bulbs but also white LEDs are not truly white. Instead, they typically are made by encapsulating a blue LED with phosphors — either a single broad yellow phosphor or a combination of green and red — which produces the desired effect when the device is excited. For general household illumination, this can be adequate; however, for a medical device, having broad color rendering and a white balance that is neither too cool blue nor too warm yellow can be critical.
However, combining red, green and blue LEDs creates a brighter white light that facilitates a broad color gamut and allows active white balancing. This makes for the ideal white-light illumination solution in many ways, but imaging the light into a small fiber bundle, liquid lightguide or diagnostic equipment optics is challenging. The traditional approach is to use a series of free-space lenses with beam-combining dichroic filters that multiplex the light onto one area. The trick is to do this efficiently with a modest amount of space and cost, while having as much of the resulting combined light as possible at a useful angle.
LED-combining optics used in the rear-projection televisions available today employ optical assemblies the size of a small shoebox, making them prohibitively large for most applications in the life sciences. This is mostly because of the size of the LED emitter area required to produce adequately bright light. Although a 12-mm2 single-emitter area sounds tiny, this is huge in the LED world, and given the predominantly Lambertian emissions, traditional approaches require large collecting optics to catch and redirect the light at a useful angle into a small area with an acceptable efficiency.
Smaller solutions are possible via molded plastic optics that rely on total internal reflection, and they can do a good job collecting light from LEDs. What sounds good in marketing literature, however, cannot quite be the whole truth in imaging applications. A claim of 90 percent collection efficiency is not much good if only 50 percent is within the f/1 or 30° half angle required for many optics systems or fiber bundles.
When using a fiber bundle, one simple solution is to split it into thirds and couple each third to an individual color. A random bundle’s combined output would appear white via simple architecture and no combining filters. A 5-mm-diameter bundle split this way would yield three 6.5-mm2-area bundles. However, using the highest-emission LEDs available still would bring into the 5-mm bundle’s acceptance angle not much better than half the light that can be achieved with larger LEDs and a filter-based approach. The trade-off to date has been to use the shoebox-sized space that an efficient lens and filter-based solution require.
An alternative to the traditional free-space approach is to contain all LED emission immediately via an enclosed column. X-cube RGB combining approaches have used this method with solid elements and immersed dichroic filters, achieving very small dimensions but small efficiencies, too. A promising novel approach employing hollow tunnels of dielectrically coated high-reflector glass can hit the “sweet spot” between LED collection efficiency and size for many applications.
Small trapezoidal assemblies of high-reflector thin-film coating can be optimized for specific collection angles from the LED, serving as an efficient means of routing light toward color-combining dichroic filters with a useful collection efficiency level that surpasses that of molded plastic collectors. An added benefit is that a hollow light pipe assembly can be added after the color multiplexing dichroic filter stage to eliminate the need for the separate light mixer rod typically required for adequate output uniformity of the light source.
Figure 3. Demultiplexing of white light is demonstrated by a pocket penlight shining “backward” through enclosed combining optics before final assembly. In the final application, these illuminated trapezoidal collection assemblies of high-reflector coated glass are where the LED emitters will be positioned.
The result is an RGB LED multiplexer with an efficiency that approaches that of large free-space optics but in a small fraction of the size, facilitating applications previously unsuitable for LEDs (Figures 3 and 4). Application-specific designs from Bookham Inc. use a patent-pending “one way” filter window between the collector and combining optics that plays a critical role in enclosed-design efficiency and is enabling LED use in a new generation of etendue-limited products.
Figure 4. Up to six wavelengths in the visible range can be multiplexed through one continuous series of dichroic filters with good efficiency if wavelength spacing is not too close. Optics sizes increase to collect light efficiently when the useful angle range for an application decreases. This optimizes dichroic combining filter performance.
LEDs for fluorescence
Another important area where LEDs have promise — but so far, limited success — is in fluorescence applications. The goal in these applications is not white but very specific narrow bands of light in or near the visible region required. Filtered bulb approaches typically are used, but besides avoiding the lifetime and heat drawbacks of bulbs, LEDs can reduce fluorescence excitation filter wide-range blocking requirements and even eliminate a filter wheel. There is a potential for the narrow emission band of an LED to be perfect for excitation and not to intrude upon a dye’s redshifted emission spectrum. Any small overlap of the exciting LED’s spectra into the measured emission band can overwhelm light produced from dyes, however.
Hard-coated, thin-film dielectric filters can be used to narrow the spectrum of the LEDs and to produce very steep edges of a few nanometers between >90 percent peak transmission and 0.0001 percent transmission. The problem is that LEDs are not always available at the right wavelengths or with enough intensity to excite fluorescence dyes properly. Furthermore, production volumes for biomedical applications that would employ these non-RGB wavelengths pale in comparison with LED applications for streetlights, for commercial or home lighting, or for consumer electronic products. As a result, it could be several years before LED manufacturers properly address the lower volumes of mere tens of thousands of high-intensity LEDs per wavelength that fluorescence applications would require.
Dichroic filters
For fluorescence applications whose wavelength requirements are well-supported by currently available LEDs, dichroic filter methods similar to RGB are usually best for multiplexing light onto microplates, although the challenges are far greater. High-performance hard-coating optical filters for excitation, emission and beam combining use alternating layers of high- and low-index material to create constructive and deconstructive interference optimized for specific wavelengths of light. As light enters these layers at an angle, the effective layer thickness is greater. When collecting light from LEDs at f/1, the dichroic combining filters designed for 45° actually combine light from 15° to 75° (Figure 5). This suboptimal situation effectively broadens the filter’s slope. If LED wavelengths are too close together, the broadening manifests itself in the clipping of adjacent wavelengths and creates lower efficiency.
Figure 5. Dichroic wavelength combining filter performance is best with perfectly collimated light. Light collection from LEDs can reduce filter efficiency as a result of the filter slope broadening. For f/1, the 45° optimized filter actually receives light at angles from 15° to 75°, clipping the LED emission spectra if the wavelengths are close. (T = transmission, HCA = half cone angle).
It is possible to multiplex five or six LED wavelengths within the visible spectrum, but only if at the right wavelength spacing. Every fluorescence dye set has its own set of requirements, but given today’s LED offerings, even the compact new space- and lumen-efficient enclosed optical approach still is limited generally to four or five wavelengths to have adequate illumination levels for a microplate. Microscopes usually can get by with less light. New LEDs are improving continually in that more light is coming from the same-size emitter area so that, unlike many areas of technology, the new versions are usually backward-compatible. This, combined with new, compact, enclosed light pipe LED collection and multiplexing methods, bodes well for further implementation of LEDs into more instrument applications.
Meet the author
Ben Standish is the product line manager for thin-film optics at Bookham Inc. in Santa Rosa, Calif.