Optical Filters for Wavelength Selection in Fluorescence Instrumentation

Turan Erdogan1

1 Semrock, Inc., Rochester, New York
Publication Name:  Current Protocols in Cytometry
Unit Number:  Unit 2.4
DOI:  10.1002/0471142956.cy0204s56
Online Posting Date:  April, 2011
GO TO THE FULL TEXT: PDF or HTML at Wiley Online Library


Fluorescence imaging and analysis techniques have become ubiquitous in life science research, and they are poised to play an equally vital role in in vitro diagnostics (IVD) in the future. Optical filters are crucial for nearly all fluorescence microscopes and instruments, not only to provide the obvious function of spectral control, but also to ensure the highest possible detection sensitivity and imaging resolution. Filters make it possible for the sample to “see” light within only the absorption band, and the detector to “see” light within only the emission band. Without filters, the detector would not be able to distinguish the desired fluorescence from scattered excitation light and autofluorescence from the sample, substrate, and other optics in the system. Today the vast majority of fluorescence instruments, including the widely popular fluorescence microscope, use thin‐film interference filters to control the spectra of the excitation and emission light. Hence, this unit emphasizes thin‐film filters. After briefly introducing different types of thin‐film filters and how they are made, the unit describes in detail different optical filter configurations in fluorescence instruments, including both single‐color and multicolor imaging systems. Several key properties of thin‐film filters, which can significantly affect optical system performance, are then described. In the final section, tunable optical filters are also addressed in a relative comparison. Curr. Protoc. Cytom. 56:2.4.1‐2.4.25. © 2011 by John Wiley & Sons, Inc.

Keywords: fluorescence; filter; optical filter; microscopy; imaging; wavelength; interference; tunable filter; thin film; dichroic beamsplitter

PDF or HTML at Wiley Online Library

Table of Contents

  • Introduction
  • Optical Thin‐Film Interference Filters
  • Optical Filter Configurations in Fluorescence Instruments
  • Fluorescence Filters Impact Optical System Performance
  • Tunable Optical Filters
  • Conclusion
  • Literature Cited
  • Figures
  • Tables
PDF or HTML at Wiley Online Library


PDF or HTML at Wiley Online Library


  •   FigureFigure 2.4.1 Fluorescence from an (A) energy and (B) wavelength perspective. In real materials, the vibrational energy levels within a given electronic state are not distinguishable, leading to broad, smooth spectra. For most common fluorophores, the vibrational energy level spacing is similar for the ground and excited states, resulting in a fluorescence spectrum that strongly resembles the mirror image of the absorption spectrum. Because of the extremely rapid nonradiative intraband relaxation, the spectral profile of the emission is approximately constant regardless of the spectrum of the excitation light. A better (poorer) overlap of the excitation light spectrum with the molecule's absorption spectrum merely increases (reduces) the total amount of emitted fluorescence.
  •   FigureFigure 2.4.2 Basic types of thin‐film interference filters used in fluorescence instrumentation. Bandpass filters are characterized by the center wavelength and the bandwidth, which may be either the characteristic full‐width‐at‐half‐maximum transmission (FWHM), or a guaranteed minimum bandwidth over which the transmission is specified. Narrow bandpass filters have a bandwidth that is a few % of the center wavelength or less, and generally comprise one or more Fabry‐Perot cavities, each of which consists of two mirrors comprised of stacks of quarter‐wavelength‐thick layers, separated by a spacer layer that is a half‐wavelength thick. Wide bandpass filters have a bandwidth that is a few % of the center wavelength or more, and are typically made by combining edge filters. Long‐wave‐pass (LWP) edge filters block (reflect) light over a wide range of wavelengths below the “cut‐on,” or edge wavelength, and then transmit light over a wide range of longer wavelengths. Short‐wave‐pass (SWP) filters are the mirror image of LWP filters and are characterized by a “cut‐off” wavelength. Because the critical function of fluorescence edge filters is blocking, which is difficult to see on a linear‐scale spectrum, edge filters are often specified in terms of the longest (shortest) wavelength that is guaranteed to be blocked by the LWP (SWP) filter. Notch filters block a narrow range of wavelengths and transmit light on either side of the blocking range. The edge steepness of a notch filter is not as steep as that of an edge filter. Dichroic beamsplitters are LWP or SWP edge filters designed to be used at a nonnormal angle of incidence (usually 45°, but sometimes a smaller angle) to either separate or combine two beams of light. Mirrors reflect incident light over a broad range of wavelengths and angles of incidence. Actual measured spectra from typical filters are shown.
  •   FigureFigure 2.4.3 Basic configurations for optical filters in the most popular fluorescence instruments. (A) Standard, three‐filter epifluorescence arrangement found in most microscopes and imaging instruments. (B) High‐reflecting spot on a window replaces the dichroic beamsplitter for systems with a small‐diameter excitation beam (like a laser). (C) Side collection (typically 90°) is useful for systems with a self‐contained sample that is accessible from all sides. (D) Forward collection is rarely used, as it is difficult to achieve sufficient blocking of scattered light in real systems.
  •   FigureFigure 2.4.4 Example (measured) filter spectra for single‐color fluorescence imaging and detection. (A) Bandpass and (B) long‐wave‐pass (LWP) emitter filter sets for imaging popular fluorescent proteins. Note the high transmission and very steep edges achievable by the modern sputtering‐based thin‐film filter technology. (C) Example of a four‐filter set for ratiometric imaging. The two exciters are designed to overlap with the saturated‐Ca2+ and calcium‐free versions of the Fura‐2 fluorophore absorption spectrum. (D) Example of a four‐filter set for monitoring the FRET interaction between CFP and YFP.
  •   FigureFigure 2.4.5 Diagrams of filter configurations for single‐color fluorescence imaging and detection. (A) The most straightforward method to swap filters is to exchange three‐filter cubes in a microscope. (B) Higher‐speed ratiometric imaging can be performed more accurately with an exciter filter wheel. FRET imaging can be done with an emitter filter wheel (C) or with an additional dichroic beamsplitter to separate the donor and acceptor fluorescence signals onto two different cameras (D) or onto different regions of a single camera for truly simultaneous, two‐color imaging.
  •   FigureFigure 2.4.6 Images of a triple‐labeled sample captured using three separate single‐band filter sets in sequence (BrightLine DAPI‐5060B, FITC‐3540B, and TXRED‐4040B from Semrock). An Olympus BX41 microscope with a 40X objective and a QImaging Retiga monochrome CCD camera were used. The sample is bovine pulmonary artery endothelial cells (Invitrogen Molecular Probes Fluo Cells no. 2), where DAPI is labeling the nucleus, F‐actin is labeled with BODIPY FL phallacidin, and the mitochondria are labeled with MitoTracker Red CMXRos.
  •   FigureFigure 2.4.7 Crosstalk illustrated. In this example, cyan and yellow fluorescent protein (CFP and YFP, respectively) are imaged from the same sample. Excitation and emission filter bands are illustrated by the solid gray bands. When only CFP is intended to be imaged with single‐band CFP excitation and emission filters, there is still a small amount of YFP absorption in the CFP exciter band, and even a small amount of YFP emission in the CFP emitter band. With a multiband emitter, as used in “full multiband” and “Pinkel” filter sets, there is a substantial amount of YFP emission since that emitter band is always present.
  •   FigureFigure 2.4.8 (A) When imaging a multicolor spot (e.g., a multilabeled microsphere), or multiple spots with different colors in very close proximity, with three separate filter sets, pixel shift (left) causes the various colors to be misaligned in the merged image at the bottom. With zero pixel shift, the spots are perfectly aligned (right), and thus here appear white. (B) In an epifluorescence‐imaging configuration, a wedge angle on the dichroic or emitter causes beam deviation (dotted path), which results in pixel shift. Wedge angles are greatly exaggerated for illustration purposes.
  •   FigureFigure 2.4.9 Example (measured) filter spectra for multicolor fluorescence imaging and detection. (A) “Full multiband” filter set capable of imaging DAPI, FITC, TRITC, and Cy5 simultaneously, for example. All filters are quad‐band filters. (B) Triple‐band “Pinkel” filter set with three single‐band exciters and triple‐band dichroic and emitter. (C) Penta‐band “Sedat” filter set with five single‐band exciters, five single‐band emitters, and a penta‐band beamsplitter.
  •   FigureFigure 2.4.10 Diagrams of filter configurations for multicolor fluorescence imaging and detection. (A) “Full multiband” configuration uses all multiband filters. (B) “Pinkel” configuration uses single‐band exciters in a filter wheel with multiband dichroic and emission filters. (C) “Sedat” configuration uses single‐band exciters and emitters in synchronized filter wheels with a multiband dichroic beamsplitter.
  •   FigureFigure 2.4.11 A complex laser‐based imaging system capable of multiple imaging modes. Different filter types are highlighted in blue. With pinholes and x‐y scanning optics active, this diagram represents a laser‐scanning confocal microscope. Multiple detection channels allow for both discrete filter detection and spectral imaging on a spectrometer (grating plus multi‐element photomultiplier tube, or PMT). The lasers could also excite the sample in TIRF mode, in which case the nondescanned path with CCD cameras could be used to collect multiple color channels simultaneously.
  •   FigureFigure 2.4.12 (A) When a beam of light reflects off of a curved mirror (or bent dichroic), the focal plane shifts relative to the plane for a perfectly flat mirror. (B) The Rayleigh range is a measure of depth of focus, and is defined for a Gaussian beam to be the length along the direction of propagation over which the beam diameter is less than times the smallest (waist) diameter, and hence appears to be “in focus.” The larger the beam diameter prior to focusing, the smaller the focal spot size (waist diameter), and the shorter the Rayleigh range.
  •   FigureFigure 2.4.13 (A) When a larger beam of light reflects off of a curved mirror (or bent dichroic), significant blurring of the spot size results from astigmatism. The best focus occurs at the “circle of least confusion” between the sagittal and tangential line foci. (B) Spot size (at the camera) that results in a microscope with a perfect 40×, 0.75 NA objective and a perfect 200‐mm focal length tube lens, when the light is reflected by an image‐splitting dichroic placed between the two lenses and with a varying radius of curvature. The diffraction‐limited spot size at the camera is about 17 µm (horizontal red line). Results are from a ray simulation using CODE V optical design software (Optical Research Associates). For this objective/tube lens combination, a radius greater than ∼50 m is sufficient to ensure the blurring caused mostly by astigmatism is smaller than that caused by diffraction. (C) Images of F‐actin in bovine pulmonary artery endothelial cells (Fluo Cells prepared slide no. 1 from Invitrogen) after reflection off of dichroics with 84‐m and 7‐m radii of curvature (Olympus BX41 microscope; 40×, 0.75 NA objective; QImaging Retiga CCD camera).
  •   FigureFigure 2.4.14 Relationship between optical density (OD) and transmission ( T) up to OD 6. The simple rules enable convenient conversion between OD and T. For example, T = 0.01 (1%) corresponds to an OD of 2; using the “÷2” rule, then T = 5 × 10–3 (0.5%) corresponds to an OD of 2.3.
  •   FigureFigure 2.4.15 (A) When incoherent light is transmitted through two coatings or filters that are immediately adjacent and parallel, the net transmission is not simply the product of the transmission values of the two filters ( TT1× T2). Therefore, the net OD is not simply the sum of the individual OD values. (B) Loss intentionally introduced between the two filters by tilting and separating them makes it possible to make the ODs approximately add. Calculated net transmission of incoherent light through two identical, parallel filters as a function of the transmission through one filter, shown on both (C) linear and (D) logarithmic scales. Note that two OD 6 filters with no loss between them result in a net OD of 6.3, whereas introduction of only 1% (or 10%) loss increases the net OD to 10 (or 11).
  •   FigureFigure 2.4.16 (A) Difference between angle of incidence (AOI) for collimated light and cone half angle (CHA). (B) Blue shift of the spectrum of a LWP edge filter as the AOI is increased from 0° to 45°. For this example, the shift of the cut‐on edge is described by neff = 2.08 and 1.62 for s‐ and p‐polarized light, respectively. Spectra at various AOI and CHA values for (C) a typical bandpass fluorescence filter and (D) a typical high‐performance dichroic beamsplitter. Note that in most microscopes the CHA in the collimated space between the objective and tube lens is 1.5° to 3°, so the dashed green curve in (D) represents realistic spectral performance of the dichroic in a larger‐field microscope. Spectra are calculated.
  •   FigureFigure 2.4.17 (A) Example of a fluorescence bandpass thin‐film filter comprised of a combination of LWP and SWP filter coatings (FWHM ∼35 nm). Six spectra associated with light of average polarization and for angles ranging from 0° to 60° are shown. (B) VersaChrome tunable filter shown for comparison. Spectra are calculated.


Literature Cited

Literature Cited
   Combs, C.A. 2010. Fluorescence microscopy: A concise guide to current imaging methods. Curr. Protoc. Neurosci. 50:2.1.1‐2.1.14.
   Conchello, J.‐A. and Lichtman, J.W. 2005. Optical sectioning microscopy. Nat. Meth. 2:920‐931.
   Erdogan, T. and Mizrahi, V. 2003. Thin‐film filters come of age. Photon Spectra 37:94‐100.
   Erdogan, T., Pradhan, A., and Mizrahi, V. 2003. Optical filters impact fluorescence fidelity. Biophot. Int. 10:38‐42.
   Galbraith, W. and Farkas, D.L. 1993. Remapping disparate images for coincidence. J. Microscopy 172:163‐176.
   Lichtman, J.W. and Conchello, J.‐A. 2005. Fluorescence microscopy. Nat. Meth. 2:910‐919.
   Moerner, W.E. 2007. New directions in single‐molecule imaging and analysis. Proc. Natl. Acad. Sci. U.S.A. 104:12596‐12602.
   Pinkel, D., Gray, J., Segraves, R., Waldman, F., Trask, B., Yu, L.C., Eastman, D., and Dean, P. 1989. New technologies in cytometry. S.P.I.E. Proc. 1063:123‐132.
   Stephens, D.J. and Allan, V.J. 2003. Light microscopy techniques for live cell imaging. Science 300:82‐86.
   Yeh, P. 1988. Optical Waves in Layered Media, Section 7.6. Wiley, New York.
PDF or HTML at Wiley Online Library