Next-Generation Thin-Film Optical Filters for Life Sciences

For any fluorescence application, three basic optical filters are needed: (1) an excitation filter where light is transmitted at wavelengths absorbed by the fluorophore; (2) an emission filter, where light is transmitted at wavelengths emitted by the fluorophore; and (3) a beamsplitter that directs the excitation light to the sample and the emission light to the detector (Figure 1).

Figure 1. Fluorescence filter cube containing hard-coated, next-generation thin-film optical filters.

Several of the emission and excitation filters are bandpass filters, which means they transmit light at a particular wavelength, or in a range of wavelengths, but block the adjacent spectrum on either sides. Beamsplitters are different and are dichroic or polychroic filters, which are set at an angle. Beamsplitters transmit certain wavelengths but reflect others.

Excitation and Emission Filter Considerations

Bandpass Design

Bandpass filters can be optimally targeted to the emission and excitation spectra of each fluorophore, or set of fluorophores, which makes them critical components of fluorescence systems (Figure 2). A bandpass filter can be designed in two different ways.

The first method involves using stacked Fabry-Perot resonant cavities, while the second integrates individual longwave-pass (LWP) and shortwave-pass (SWP) filters to form both edges of the bandpass. Cavity filters are inherently symmetric in shape, while LWP and SWP can be designed in an asymmetric manner to lower cost in some situations.

Figure 2. Excitation and emission spectra of SYBR^® Green I DNA binding stain overlaid with the transmission spectra of a set of optical filters designed to optimize detection and visualization of this fluorophore.

To block light transmission outside of the passband, or high transmission region, the two types of bandpass filters are paired with blocking elements. Out-of-band blocking is generally described by the optical density (OD) of the filter, which is defined as the negative log (base 10) of the transmittance (T).

Blocking elements can be dielectric blockers such as additional LWP and SWP filters, or non-dielectric elements including light-absorbing colored glass, which is expensive, introduces autofluorescence, and has reliability concerns.

Figure 3 shows a typical transmission response (%T), as a function of the electromagnetic spectrum, for various common emission and excitation filter bandwidths and wavelengths. These filters are completely blocked, meaning all wavelengths outside the passband are rejected, to OD6 or greater, so that optimized S/N ratios can be ensured.

Figure 3. Typical transmission levels of several different excitation and emission bandpass filters. High transmission is able to be achieved across the electromagnetic spectrum from UV to IR wavelengths. All filters are blocked to a level of OD6 or greater.

Although soft-coated thin-film bandpass filters are available, hard-coating alternatives have better durability, and offer higher in-band transmission and steeper transitions to high out-of-band (OD) blocking. These factors improve S/N in fluorescence systems, which can also improve outcomes for laser scanning fluorescence confocal microscopy, in which S/N decides the achievable contrast of the final image.

Slope and Squareness

A signal’s “squareness” is also critical as it enables a steeper transition between high %T in the passband to high OD blocking. The edges of the bandpass (or edge) filter are more vertical if the filter shape is squarer. This might be critical in Raman spectroscopy, where the laser line filter must have a steep cut-off edge to modulate the laser source.

Conversely, the laser blocking filter, which is often a Raman longpass, must have a steep cut-on edge to completely block the light from the laser. Designing bandpass filters with improved squareness or steeper slopes can be easily achieved by adding more cavities to the filter (Figure 4).

Improving the slope of the LWP/SWP designs is almost the same, although here, it involves addition of more layers.

Figure 4. Filter “squareness” is a direct function of the number of resonant Fabry-Perot cavities in cavity filters, or the number of layers in LWP/SWP designs.

However, cavity additions can lead to undesirable ripple and loss to the passband, which results in contrast reduction in the final image. Fortunately, a unique optical thickness monitoring technique enables increased cavity counts while a low passband ripple is maintained.

The system works by continuous measurement of filter function and compensating for thickness errors associated with prior layers. Figure 5 shows how low ripple and high transmission were repeated by using this technique for four coating runs of the same high-cavity bandpass filter design from Figure 4.

The monitoring technique enables filters to satisfy high-performance system requirements and not compromising accuracy or precision.

Figure 5. Repeatability of high cavity count filters. Shown above are four runs of fully blocked high cavity count bandpass filters from three different chambers.

Out-of-Band Blocking

One of the characteristics of fluorescence detection filters is the blocking level, especially with respect to the overlap of the excitation filter. Out-of-band blocking levels of 5 to 6 OD are preferred for several fluorescence applications. However, care needs to be taken at wavelength ranges or crossover points where excitation response reduces and emission response inclines.

In several cases, blocking coverage needs to be more at these points as inadequate blocking in this region can lead to undesirable light leakage. For instance, if the emission filter is not able to sufficiently block the much brighter excitation light, it will surpass the weaker emitted fluorescence light from the specimen, resulting in poor S/N.

Figure 6 illustrates the blocking levels of the bandpass filter depicted in Figure 5. This filter features the bandpass and blocking layers on one side of the substrate, improving transmission and reducing cost. An anti-reflection (AR) coating on the back of the substrate enhances transmission at nominal extra cost.

Figure 6. Repeatability of blocking levels for the four runs of the high cavity count bandpass filter seen in Figure 5. All are blocked to a level of OD6.

Multi-Bandpass Filters

The flexibility of multiple bands of illumination for excitation and emission is a requirement for the latest bioimaging systems. Next generation thin film hard coatings resolve such demanding applications, and deliver high transmission and low ripple for multiple passbands, as shown in Figure 7. Multiband filters are usually specified in terms of blocking (rejection) bands and transmission bands.

Figure 7. Excitation and emission spectra for both EYFP and MitoTracker^® Red overlaid with the transmission spectra of a dual-band filter set designed for these fluorophores.

90 - 95% transmission levels are common within the passband, while average blocking levels are from 5OD and 6OD between bands. In multiband filters, the gaps between the passband and blocking bands are usually 2 - 3% of wavelength, although cost increases if the spectral gaps are narrower.

Cost can be reduced by avoiding specifications such as >6OD blocking levels and steep transition slopes, unless they are necessary for the system.

Beamsplitter Considerations

Dichroic and Polychroic Filters

A dichroic filter separates the spectral bands of the excitation and emission light by minimizing signal loss and spectral gap between bands, as shown in Figure 8. Improvements in image contrast for many applications such as fluorescence and multispectral imaging can be enabled by sharp transitions from reflection to transmission.

For implementation in these systems, dichroic filters must have low angle shift, excellent flatness, and low polarization splitting of both reflected and transmitted wavefront to remove any noticeable focal shift, which is detrimental for any imaging system. This is particularly critical for total internal reflection fluorescence (TIRF) and structured illumination microscopy and other applications.

Figure 8. Transmission of a dichroic beamsplitter measured at a 45° angle.

Multiband systems have better performance levels, and need a comb-like performance that seperates multiple excitation bands from multiple emission bands. These types of filters enable the study of cells in real-time. This is particularly important in the field of cancer research for real-time analysis of stem cell growth. Figure 9 demonstrates the spectra for a polychroic (multiple wavelengths) beamsplitter.

Figure 9. Comparison of measured vs. theoretical spectra for a quad-band polychroic beamsplitter. Transmission levels were measured at a 45° angle.

General Filter Considerations

Coating Stress

Deposition of a coating usually creates stress inside the thin film, which often leads to alteration in the flatness of the optical substrate. The type of stress, compressive or tensile, is associated with the type of coating process and the details of the coating conditions.

Generally, coating stress leads to a slight curvature of the substrate, forming a hill or bowl shape. An optical interferometer can be used to measure this curvature as shown in Figures 10 and 11.

Figures 10 (top) and 11 (bottom). Curvature of substrate due to coating stress. The figure on the left shows the measured phase obtained using an interferometer while on the right is the measured optical path difference as compared to a reference flat.

Curvature effects on an optical beam or image is somewhat similar to that of a lens, and may introduce undesirable distortion into the optical system. This may become vital, for instance, when the accurate position of many wells containing fluorescently-tagged DNA needs to be measured from an image produced by a CCD detector. Accurate measurements can only be made when the curvature effects are minimized.

Several techniques can be used to compensate stress-induced curvature. One of the simplest ways is to increase the substrate thickness because the use of additional strategies can increase the filter cost. Therefore, maintaining a low level repeatable stress is necessary to keep costs low and yields high.

Variations in parallelism, thickness, or index of refraction across the filter can also lead to distortion of transmitted light. This type of distortion, called transmitted wavefront error (TWE), can also be determined using an interferometer, as shown in Figures 12 and 13.

Improvements in instrumentation design have driven the levels of TWE performance such that measureable amounts of TWE related to the coating need to be carefully managed to optimize image quality.

Figures 12 (top) and 13 (bottom). TWE, or change in optical path length in transmission. The figure on the left shows the measured phase using an interferometer while the right shows the variation in optical path length through the part.

Summary

Arguably the most important elements defining system performance are the optical filters integrated into fluorescence imaging and detection systems. Next-generation thin-film optical filters feature improved transmission, blocking, and transmitted and reflected wavefront properties.

This leads to bright, high-contrast images and accurate detection of target molecules. Alluxa provides these high-performance filters at competitive prices and at any production volume. For any fluorescence-based instrument, its team of experts can help produce custom filters in order to achieve optimized system performance.

This information has been sourced, reviewed and adapted from materials provided by Alluxa.

For more information on this source, please visit Alluxa.