Making a Basic Illuminator
LEDs can be effectively used as illuminators in a standard microscope. The simplest case is the one where only a single illuminating wavelength is required. For this purpose, one can match and even exceed the illumination provided by Xenon lamps through either of two simple coupling methods to be discussed later. The main concerns here are adequate cooling, and importantly, use of a DC regulated power supply.
Mounting, power supply and circuitry
Cooling can be achieved by a large surface area fin type heat sink as shown in , commonly used for CPUs in computers, with additional use of a fan if the LED is to be used in a small confined space. The LED and the heat sink should be thermally connected using thermal grease available from any electronic supply house and a thermally conductive electrical insulator (e.g. mica). Although LEDs generate far less heat than arc lamps (normalized to the light output), temperature-dependence of emission necessitates careful experimental design when LED power is turned on and off as a way of shuttering image acquisition.
The power supply must be chosen with care to allow stable illumination, as well as rapid modulation of intensity (even if it is simply turning light on and off). Illustrated in is a sample voltage driver circuit for controlling a high intensity blue Luxeon LED with a simple low current 5 V TTL input as the switching signal. This signal can be obtained from an IO board, the serial port, or the parallel port on a computer. The light intensity is controlled by the potentiometer. Care must be taken to stay within the safe maximum driving current (usually specified by the manufacturer). Unfortunately, LEDs of the same model can vary substantially (~1 V) in their forward voltage drop (Vf) due to differences in manufacturing and inhomogeneities in the raw building materials. This translates into differences in the current that different LEDs from the same “batch” pass at a particular holding voltage. Therefore, to avoid damaging the LEDs, when using a voltage source, one should determine in advance the holding voltage to current relationship for any given LED unit. Then, simply adjust the potentiometer or the holding voltage of the controlling circuit accordingly so that the maximum recommended current is not exceeded (). Alternatively, to maximize the LEDs performance, inexpensive current regulated LED power supplies can be used (for example, 3021/3023 “BuckPuck” driver, wiring diagrams and datasheets can be found at:
http://www.leddynamics.com/LuxDrive/datasheets/3021-BuckPuck.pdf, or the MAX16818 current driver from
http://www.maxim-ic.com/.
The responses of an LED driven by a minimal electronic circuit consisting of a single transistor and 0.5Ω resistor are shown in . We recorded the light intensity using a photomultiplier tube (R3896+C270 DAP socket, Hamamatsu Photonics, K.K., Hamamatsu City, Japan) and an IO DAQ board (PCI6110E, National Instruments Corporation, Austin, TX). Not surprisingly, LEDs driven at high rates (10 KHz) can follow these changes well. The on/off kinetics appears to be dominated by time constants of electronic components from the driving circuit of choice – it is around 10 µs in this present example with a rather simple, suboptimal circuit.
Coupling the light source to the microscope
The LED should be coupled to a microscope or used to illuminate the specimen directly. A few schemes for coupling of the light source are described and illustrated below. While designing coupling optics, a key consideration is the so-called Lagrangian invariant, or more precisely a related quantity called étendue. Étendue is proportional to the product of the source size and the divergence angle of the emission, and provides an estimate of the maximum amount of light that can be collected through an aperture. Proper acknowledgement of this invariant is useful in avoiding unnecessary or impossible coupling. The reader can find a thorough discussion of invariants in Born and Wolf
[12], or a more introductory discussion in Fischer and Tadic-Galeb
[13].
One can use a light guide, such as Edmund Optics's Liquid Light Guides (NT53-691, Edmund Optics, Inc., Barrington, NJ) to transfer the light and shine it onto a wide area or to direct the light from the LED to a desired location such as to a microscope (). The placement of the optical fiber or light guide in relation to the LED is important to consider. For surface emitting LEDs whose emitting surface is the same as the fiber size, one can simply place the light guide as close to the surface of the LED as possible (butt-coupling). If the fiber size is smaller, then some coupling optics can be used to match the size of the fiber to the source. But this will only help when the étendue of the fiber is greater than or equal to that of the source. A nice discussion of these issues can be found in an article by Doric and Tubic in the October 2004 issue of Laser Focus World (reprint also found in
http://www.doriclenses.com/lire/39.html). For all practical purposes, the butt-coupling approach allows for the best light collection and dispenses with unnecessary and expensive optics.
Since the excitation and emission wavelengths are quite close for some fluorophores, an excitation filter should be used with the LED. In this case, the filter could be placed either directly on the LED surface, possibly sandwiched between the LED and the light guide, be mounted at the other end of the fiber, or in a traditional filter wheel if a microscope is to be used. Alternately, one might also rely on the existing filter cubes in the microscope if they match the emission wavelength of the LED arrays. The exact configuration depends on whether the light source is coupled to a microscope or used as a standalone illumination source.
shows a simple method to couple LED lighting sources to an Olympus BX51WI fixed stage microscope (Olympus America, Inc., Center Valley, PA). A high intensity blue LED from Lumileds (Luxeon LXHL-LB5C peak emission at 470 nm) was coupled to a 1 mm optical fiber (BFH48-1000 Thorlabs, Inc., Newton, NJ) Using the fiber optics, light from the LED is sent to the microscope into one of the slots meant for the neutral density filters as shown in the figure. There, a mirror is used to align the light onto the optical path that is normally used by a xenon lamp attached to the the microscope via a light guide (Polychrome V from Till Photonics). If an excitation filter was not used on the LED, one can be placed in the microscope at this location or in the filter wheel.
In the alternate coupling method, shown in , the same LED and driving circuitry are attached to a tube and lens, and coupled to the microscope through the rear lamp housing port. The tube length and the lens should be chosen specifically so that the principles of Köhler illumination are obeyed. This will ensure homogeneous illumination of the specimen. In practice, this involves mounting the tube such that it can be translated smoothly to allow focusing of the LED and following procedures similar to those used for arc lamps. A less effective, but simpler way of reducing inhomogeneity involves the usage of a diffuser placed between the LED and the collection lens. With this combination, the tube position can be adjusted to achieve maximum brightness. Any residual inhomogeneity in illumination can be corrected (in principle) by applying flat field correction digitally (methods can be found in standard optical microscopy or image processing text books). In our laboratories, we have used lens (LA1805-A), tube (SM1L30) and diffuser (DG10-600) from Thorlabs.
If the LED is to be used as a standalone illumination source (rather than through a commercial microscope's optical relay), one simply needs to consider whether to have focusing or expanding lenses attached at the end of the light guide and whether to include an excitation filter in the light path.
Two Wavelength Illuminator
When more than one excitation wavelength is desired, independent electronic circuits can be constructed to drive multiple LED arrays similar to that shown in , along with a small optical relay to couple two different LEDs onto a single optical fiber bundle. While it is possible to adjust the intensity of the LEDs from the software using pulse width modulation, for most experiments requiring simply an on or off state for each LED, the intensity can be much more easily regulated in this circuit with a potentiometer.
The optical setup shown in has two LEDs with appropriate excitation filters mounted (D480/20, Chroma Technology Corporation, Rockingham, VT). The output light is focused and reflected off a dichroic mirror that allows passage of one wavelength (780 nm), while reflecting the other (470 nm). Both light sources are thus aligned onto the edge of the light guide or optical fiber used.
Comparison of arc lamps with LEDs
We compared fluorescent images of fixed tissue sections labeled with antibodies against the astrocytic marker GFAP, obtained sequentially using an LED driven at 700 mA and a xenon-arc lamp (TILL Polychrome, TILL Photonics, LLC., Pleasanton, CA). The LED was coupled to the microscope by simply placing one end of an optical fiber on the LED chip. The dichroic, objective lens and emission filter were the same for both light sources, as was the exposure time of 25 ms. The images and intensity plots in illustrate that, for matching imaging conditions, the LED signal was 30% brighter and as uniformly focused on the sample as the xenon-arc signal. Improvements in light coupling, transmission, and LED technology will likely lead to even brighter LED based light sources.
A similar comparison with a 100 W mercury lamp revealed that the output power of an LED array at the focal plane of the objective was approximately ten times less than that of a mercury lamp. Thus, in terms of available output, top end LEDs currently cover a range of intensities between the xenon and mercury lamps. Their output power is however sufficient for most biological applications, including imaging of weakly fluorescent probes like synaptopHluorins (see below).
Custom-built widefield microscope for in vivo imaging
In this final example, we present a sample application in which real-time fluorescence imaging can be done in an intact neural system using LED illumination. The preparation is the mouse olfactory bulb
[14]. Each glomerulus in the adult olfactory bulb receives input from a single type of odorant receptor (~1300 types in mice). Within the glomerulus, tens of thousands of sensory axons make synapses on principal cells and interneurons. Glomeruli are located a variable distance below the surface of the bulb, ranging from 10 µm to a 100 µm; each glomerulus is about 80 µm in diameter in the mouse. Odorants bind to receptors in the nose and activate sensory neurons, which then release neurotransmitter in the glomerulus. Thus each glomerulus on the surface of the bulb has a specific chemical response spectrum, derived from the associated odor receptor.
Two optical signals are used to measure responses in glomeruli - intrinsic optical signals, derived from changes in near infrared light scattering
[15]–
[17], and fluorescence signals from either calcium indicators
[18] or the presynaptic probe synaptopHluorin (SpH)
[14],
[19]. Here we show measurements of intrinsic optical signals and synaptopHluorin signals.
To image olfactory bulbs, we coupled two lenses (Nikon Telephoto 105 mm f/1.8 AIS Manual Focus Lens and Nikon Normal 50 mm f/1.2 AIS Manual Focus Lens, Nikon Corporation, Melville, NY) using a 62 mm to 52 mm stepdown ring and a 52 mm to 52 mm coupling ring. This system of lenses was attached to a CCD camera (VDS Vosskuhler CCD-1300F, VDS Vosskuhler GmbH, Osnabruck, Germany) using an F to C coupler. An optional set of extension tubes can be used between the F to C coupler and the 62 mm lens to adjust magnification. This optical relay was chosen to facilitate imaging over a large area and depth of field, minimize light loss, and to provide flexibility with different lenses. A standard macro lens could be used for the same purpose, but is more expensive and less flexible than the lenses we use. To allow fast switching between two wavelengths, we assembled the box depicted in that would hold LEDs of two wavelengths with an appropriate dichroic mirror to direct light to a quartz fiber optic light guide (NT38-956 Edmund Optics). To collect signals in the wavelength range from green to near infrared, we used a longpass filter (HQ510LP, Chroma Technology). Photographs and schematics of the device are shown in .
Imaging was performed as in Meister and Bonhoeffer
[16]. Mice were anesthetized with a ketamine/xylazine mixture. A craniotomy was performed in which the skull over the olfactory bulb was removed. A 1% agarose solution and glass coverslip were placed on top to support the bulb and prevent damage. Either blue or red light from an LED array was directed over the olfactory bulb using flexible fiber optic bundles (). Images of green fluorescence and red backscattered light were acquired using a CCD camera. illustrates the baseline signals obtained under infrared and blue lights, along with corresponding intrinsic and spH responses evoked by odor applications. Odors were delivered using a custom-built olfactometer
[16]. The blue and infrared LED arrays were toggled on and off sequentially, such that both intrinsic and fluorescence signals could be obtained in the same individual trial. Intrinsic signals were passed through a spatial filter kernel as described in Meister and Bonhoffer
[16] to isolate the high spatial frequencies corresponding to individual glomeruli odor triggered responses. SpH fluorescence was directly analyzed without further processing. As illustrated in , odor-evoked changes in both fluorescence and intrinsic optical signals can be easily detected using LEDs as an illumination source. These experiments demonstrate the feasibility of quantitative real-time imaging in a relatively complex biological preparation.
This article has been 
=