WYTZE E. VAN DER VEER AND DON WOLPERT
Many microscopy techniques require some of the wavelengths of light being observed to be suppressed. For instance, in fluorescence microscopy, to aid in detecting the emission from the sample, the excitation wavelength must be suppressed. Even more stringent requirements are encountered in two-photon and other nonlinear schemes. Eliminating the undesired wavelengths is a seemingly simple task that just requires buying a filter with the right wavelength response and mounting it in the optical pathway. However, some filter materials show strong fluorescence features that can make it necessary to use these filters with great care.
Basic Fabry-Perot filters are often used to form transmission-band filters, which can have a pass band as narrow as 3 nm. A classical Fabry-Perot filter is constructed from one or multiple quarter-wavelength-stack reflectors coupled by a thin-film layer that is an integral number of half the central wavelength of the filter. Besides the transmission peak, these filters show out-of-band transmission. As a rule of thumb, blocking by the interference coating will be effective on the short-wavelength side to 0.8 times the central wavelength of the filter and on the long-wavelength side to 1.2 times the central wavelength. A detector with a wide spectral response placed behind such a narrow-bandpass filter could thus possibly “see” wavelengths shorter and longer than the central wavelength.
For the basic interference filters, the rejection is limited to on the order of 10—4. Blocking of the out-of-band transmission of these filters can be accomplished by adding additional dielectric reflective blocking layers. More commonly, the rejection characteristics of these filters are assisted by depositing the interference filters on a colored-glass filter as a substrate. In particular, this can remove the out-of-band transmission features on the blue side of the central wavelength.
Colored-glass filters are almost ideal optical elements: they are rugged, inexpensive, and show good optical transmission, while dramatically suppressing the absorbed wave (in some cases, to a ratio of greater than 106). However, the benefit of colored glasses does not come for free, as they can be excited by the same short-wavelength light they are intended to block, generating fluorescence. The magnitude of this fluorescence is dependent on the glass type, the wavelengths of excitation, and the intensity of the excitation. The end effect is also dependant on how the filter is used in the optical system.
In our work we encountered serious problems resulting from this effect, so we measured the fluorescence of a few selected colored glasses; results of these tests are presented here.
How filter glass fluoresces
Colored filter glasses such as those obtained from Schott (Mainz, Germany) are manufactured by adding semiconducting materials to the glass melt. These materials absorb photons that have an energy larger than the bandgap, but they do not interact with photons having less energy than the bandgap. The position of the cut-off wavelength of such filters can be adjusted by choosing different semiconducting additives. These longwave-pass filters absorb at wavelengths shorter than the cut-on wavelength and pass longer wavelengths. They can be used to accomplish most of the required short-wavelength blocking.
When light is absorbed in one of these glasses, an electron in the semiconducting material is promoted to the conduction band (see Fig. 1). If all goes well, the energy now held by the electron is removed by nonradiative processes (phonon relaxation and diffusion). Although it is drawn in the figure as a single arrow, this can be a cascade of multiple steps. The excitation energy can also be optically re-emitted. Typically, the electron will first lose the excess excitation energy above the bandgap in a very fast nonradiative process. Next, the population at the bottom of the conduction band will be available for spontaneous fluorescent decay. The electron can access any state within the valence band, giving a broad spectral distribution to the generated light. The wavelength of the maximum intensity of this distribution is expected to be close to the “cut-off” wavelength for optical absorption. Also, the shape of the emission spectrum is not expected to vary as a function of the excitation wavelength.
Measuring the fluorescence
We measured the fluorescent intensities of a few selected colored-glass samples, Schott OG 515, OG 530, and RG 610, with a Hitachi F4500 fluorescence spectrometer (see table). This instrument is normally used for measuring the fluorescent characteristics of liquid samples. To accommodate glass samples, the holder for a 10 × 10 mm sample cuvette was replaced with an aperture bracket to position the colored-glass samples (see Fig. 2). The same excitation and emission geometry was maintained by placing the glass samples at 45° to the illumination beam.
To obtain the absolute fluorescent yield of these filters, dye solutions of coumarin 500 (CM500) and rhodamine 6G (R6G), which have known quantum efficiencies, were prepared.1, 2 The emission peak of CM500 at 540 nm is close to the fluorescence maxima of the Schott glasses OG 515 and OG 530. Likewise, the emission peak of R6G is close to the fluorescence maximum of RG 610 glass. The dyes were placed in a thin fused-silica sample cell with a liquid path length of 1 mm.
The absorption of the investigated glass filters was very large (optical density greater than five). To make an optimal comparison, the excitation wavelength and concentration of the dyes were chosen to obtain an equally large absorption. For this purpose, the CM500 dye was dissolved in methanol obtaining an optical density of 3.5 at 390 nm. The literature value for the quantum efficiency of this dye solution is 68%. Likewise, the R6G dye was dissolved in ethanol to obtain an optical density of 2.9 for the 1 mm cuvette at 488 nm. The literature value of the quantum efficiency of this solution is 79%.
To mimic the dye cell, the glass filters with a thickness of 1.5 mm were sandwiched between two microscope slides, each 1 mm thick. In doing this, the total thickness of the samples is identical to that of the dye cell. One difference is that the glass sample sandwich contains a total of six air-to-glass transitions. In contrast, the dye cell only has two fused-silica-to-air transitions and two fused-silica-to-solvent transitions. The measured values are corrected for these reflection losses. The fluorescence from the dyes is much brighter than that of the filters, and to scale the intensity of the fluorescent dyes to the same magnitude as the fluorescence of the glass filters, neutral density filters are used. These filters are placed in the excitation beam so that the scattered light is reduced to the same level as the fluorescence.
Experimental results
In our test setup, the sample cell containing the fluorescent dye is set at 45° to both the incident light beam and the monochromator that is measuring fluorescence. In this configuration the system is susceptible to scattered light at the point where the colored-glass samples start to become transparent. In the three-dimensional fluorescence plot of OG 515, the scattered-light component is clearly visible as a sharp ridge (see Fig. 3). The fluorescence spectrum of OG 515 is shown with the excitation wavelengths plotted versus the emission wavelengths, with the third axis being relative intensity. All the glasses we measured began fluorescing when illuminated at the shortest wavelengths up to where the glasses begin transmitting. The fluorescence is thus induced throughout the full absorption spectrum of the particular glass being illuminated. The absolute measured fluorescent intensities of the filter glasses were measured.
Conclusions
Colored glasses can be used with narrowband optical filters to remove unwanted short-wavelength light from a microscope system. The suppression can be very large (optical density greater than six) but these filters can also show strong fluorescence. The fluorescence maximum is located close to the cut-off wavelength of the glass. The intensity of the recorded signals from the glasses was surprisingly large. Especially for OG 515, the fluorescence is a very intense 9.7%. Great care should be taken using this filter.
As opposed to the light carrying the image, the fluorescence generated by the filter is distributed equally over all directions. In practice this means that to avoid artifacts, one should place these filters as far as possible from the imaging planes in the microscope.
REFERENCES
- L.S. Rohwer and J.E. Martin, J. Luminescence, 115, 77 (2005).
- U. Brackmann, Lambda Chrome Laser Dyes, second edition, Lambda Physik, Gö ttingen, Germany 1997.
WYTZE E. VAN DER VEER is the director of the Laser Spectroscopy Labs at the University of California, Irvine; e-mail: [email protected]. DON WOLPERT is a consultant with Bio-Optics, Los Angeles, CA; when this work was done, Wolpert was with Northrop Grumman, Redondo Beach, CA.