Tunable Bandpass Filters

The ubiquitous techniques of fluorescence and Raman imaging and spectroscopy rely heavily on spectrally precise, high-quality and high-throughput optical filter technologies. As both fluorescence and Raman-based techniques move from traditional R&D-based environments into medical and clinical (diagnostic) settings, even higher demands are placed on system performance. Therefore, it is necessary to continue to improve system components and architectures to meet the demanding challenges often encountered in biological applications.

Figure 1. Illustration of tunable bandpass (center) and edge (right) filters resulting from angle-tuning.

One simple and straightforward means to do this without a complete system redesign and overhaul is to conceive of new spectrally precise, high-quality and high-throughput optical filters that can perform in more than one mode of operation. A recent advance in this area has been the development of thin-film interference filters that are tunable over a wide range of wavelengths with little or no sacrifice in filter performance, i.e. transmission characteristics. This innovation is not only simpler than alternative approaches (e.g. liquid crystal technologies) but can lead to cost reductions as fewer filters are required to cover a broad wavelength range. The advent of this new technology serves to improve and expand the flexibility and capability of both fluorescence and Raman instrumentation/modalities across a wide range of imaging and spectroscopy applications.

Thin-film filters are the ideal solution for wavelength selection in most optical systems due to their exceptionally high transmission (close to 100%), very steep spectral edges, and blocking of optical density 6 or higher over wide spectral regions for maximum noise suppression. However, until now thin-film filters have been considered “fixed” such that changing the spectral characteristics required swapping filters. Mechanical means to perform filter swapping, like filter wheels, exist but these are generally large in size, relatively slow (minimum switching times are typically 50 to 100 ms), and permit only a limited number of filters. (Typically, filter wheels can contain anywhere from 4 to 12 filters, depending on the instrument and application.) As a result, size, speed, and filtering function flexibility are all limited. Therefore, in order to overcome these shortcomings new innovations in filter technology are required.

Angle-Tuned Thin-Film Filters

Figure 2. Example of a fluorescence bandpass thin-film filter comprised of a combination of long-wave-pass and short-wave-pass filter coatings (FWHM ~ 35 nm). Six spectra associated with light of average polarization and for angles ranging from 0° to 60° are shown. (b) Transmission spectra for a green-to-blue tunable thin-film filter shown for comparison. All spectra shown are calculated.

It is well-known that the spectrum of any thin-film filter shifts toward shorter wavelengths when the angle of incidence (AOI) is increased from 0° to larger angles *. However, in general the spectrum becomes highly distorted at larger angles, and the shift can be significantly different for both s- and p-polarized light, leading to strong polarization dependence.

Mathematically, when the AOI of light impinging upon the filter is increased beyond 0° (normal incidence) to larger angles, the resulting wavelength shift is generally described quite accurately by the equation where θ is the angle of incidence and neff is called the “effective index of refraction,” which is unique for each filter design and for the two orthogonal states of polarization. This effect can be used to tune the spectrum of an optical filter, albeit over a limited spectral range.

Multi-cavity Fabry-Perot thin-film filters are one example of tunable filter technology. However, although they can be designed to provide a narrow passband (about 2nm) at 561nm, the passband can become considerably narrower for s-polarizations and wider for p-polarization, and tune at different rates resulting in polarization splitting as the AOI is increased. Such polarization dependent features are undesirable and can severely limit filter and system performance in many applications.

Figure 3. (a) Transmission spectra at four AOIs for both s- and p-polarization light for a green-to-blue filter. Note the almost nonexistent polarization dependence. (b) Shift of center wavelength with increasing angle of incidence (AOI) for the same filter design; the full-width-at-half-maximum (FWHM) bandwidth remains fixed at 20 nm.

Multi-cavity Fabry-Perot thin-film filters are one special case. Many fluorescence imaging and quantization applications require filters with passbands that are considerably wider — often 30 to 50nm or more (at visible wavelengths). One approach is to form such filters using a combination of long-wave and shortwave pass edge filters. Such filters exhibit qualitatively similar behavior to multi-cavity Fabry-Perot filters in that distortion of the filter spectrum increases as the AOI is increased. The useful tuning range for such filters is typically limited to 10° to 15°, resulting in a wavelength tuning range of only 0.5-1.0%.

The above examples stress the importance of the polarization state of light on tunable filter performance. However, for many applications unpolarized light is used. Presented in Figure 2 are two transmission spectra for average polarization which clearly illustrate this aspect. Figure 2(a) shows the average polarization spectra at six angles (ranging from 0° to 60°) for a fluorescence filter. The spectrum is highly distorted (i.e. loss of steep edges) even at angles of 20° to 30°, and almost unusable for larger angles (30° to 60°).

In order to overcome these shortcomings it is necessary to develop new filter technologies with advanced designs that maintain steep (bandpass) edges, high transmission, and out-of-band blocking with essentially no or little polarization dependence. The introduction of new (proprietary) filter technology has led to the development and introduction of both bandpass and edge filters that overcome all of the shortcomings discussed above. An example of the filter performance of this new filter type is displayed in Figure 2(b). In contrast to the spectrum of the bandpass filter shown in Figure 2(a), this new innovative design results in high transmission, steep edges, and excellent out-of-band blocking over the full range of angles from 0° to 60°.

In addition to maintaining the above desired filter attributes, this new filter technology also solves the problem of polarization-dependent spectral distortion. The plot shown in Figure 3(a) shows a series of transmission spectra of a green-to-blue filter and clearly demonstrates improved transmission fidelity even at high tuning angles for both sand p-polarized light. Furthermore, from the plot displayed in Fig. 3(b) it is clear that the FWHM bandwidth, rather than narrowing, remains near constant as the AOI is varied from 0° to 60°. Continual design improvements will likely lead to even narrower passbands that are tunable over a much wider wavelength range.

Conclusion

Fluorescence microscopy and other fluorescence imaging and quantitation applications, hyperspectral imaging, high-throughput spectroscopy, and fiber optic telecommunications systems should all benefit from tunable optical filters with the spectral and two-dimensional imaging performance characteristics of thin-film filters and the center wavelength tuning flexibility of a diffraction grating. There exist several technologies that combine some of these characteristics, including liquid-crystal tunable filters, acousto-optic tunable filters, and linear-variable filters, but none are ideal and all have significant additional limitations. The introduction of tunable thin-film filters overcomes the shortcomings of alternative technologies and should pave the way for their introduction into fluorescence and Raman spectroscopy systems, leading to enhancement of the size, speed and filtering function in addition to expanding current fast-imaging capabilities.

This article was written by Neil Anderson, Ph.D., Technology Development Analyst, and Turan Erdogan, Ph.D., co-founder and CTO, Semrock, Inc. (Rochester, NY). For more information, contact Dr. Anderson at nanderson@ idexcorp.com, Dr. Erdogan at terdogan@ idexcopr.com, or visit http://info.hotims.com/28055-201 .

References

* Optical Waves in Layered Media, P. Yeh, Wiley, New York, 1988, Section 7.6