Exchanging ions with a low polarizability and small size (e.g., ) with ions of high polarizability and large size in a glass slide leads to a region of high refractive index that can serve as an optical waveguide.1 In 1972, Izawa and Nakagome2 first reported on ion-exchanged waveguides by in silicate glass containing oxides of sodium and potassium. Since then, ion-exchanged waveguides have been fabricated using various ions, such as , , , , , , and ; salts, such as , , CsCl, , and ; salt mixtures such as , , , and , (Ref. 3); and several different glasses, such as soda-lime, BGG21, borosilicate, Schott BK7, aluminisilicate, and Corning 0211 (Refs. 1, 4, 5).
Waveguides fabricated by ion exchange have taken on an important role in many applications, such as sensors,6, 7 nonlinear integrated optical switches,8 and evanescent scattering microscopy.9 In the present study, we introduce a glass that shows a very high diffusion constant for Ag ions and apply the fabricated waveguides to waveguide evanescent field fluorescence (WEFF) microscopy.9, 10, 11, 12, 13, 14 This is an inexpensive alternative to total internal reflection fluorescence (TIRF) microscopy.15 Images of cell–substratum contact regions of HEK293 cells were obtained using WEFF microscopy and compared to images obtained using classical TIRF microscopy.
We have established an ion-exchange protocol to fabricate waveguides with a defined number of modes. With the help of -line spectroscopy,16, 17 we characterized the fabricated waveguides with respect to the effective refractive indices, the effects of exchange-time and temperature, and the diffusion profiles and the temperature-dependent diffusion coefficient of the silver ions. The penetration depths of the evanescent field for different modes were calculated. The integral of the intensity of the evanescent field was investigated for transverse electric (TE) and transverse magnetic (TM) modes. ions, Ag aggregates, and nanoparticles can be present in such ion-exchanged waveguides and can lead to photoluminescence.18, 19, 20, 21, 22, 23, 24 Therefore, fluorescence emission spectra excited at different laser lines were recorded. Usually in fluorescence microscopy, background signals such as autofluorescence of the substrate are avoided, but in the case of waveguide microscopy, they can be used as an internal standard for normalizing the coupling efficiency during analysis of intensity data.24
Glass slides of SG11 (SCHOTT, Grünenplan, Germany) with a thickness of cut into pieces were first sonicated in 2% Hellmanex (Hellma, Berlin, Germany) solution for and then rinsed 20 times with Milli-Q water (Milli-Q, , Millipore) and sonicated in Milli-Q water for . Next, the slides were rinsed 20 times with ethanol (anhydrous, Sigma Aldrich), sonicated in ethanol for , and dried with nitrogen gas. The cleaned glass slides were placed on top of molten salt in a custom-made glass spoon at a fixed temperature in a tube furnace (Yokogawa M&C Corporation, Tokyo, Japan). This allowed ion exchange to happen only at the interface of the glass and molten salt. The ion-exchange temperature and the diffusion time were varied from and from to hours, respectively.
After the ion exchange, the glass was removed from the melt, cooled with nitrogen gas under ambient conditions, thoroughly rinsed with Milli-Q water, and sonicated for in Milli-Q water. Cooling the glass slides with nitrogen gas helped to control the ion-exchange time, especially when the time is small.
-Line Spectroscopy with Prism Coupling
Waveguide coupling was performed using a home-built prism coupling setup in -line geometry. The prism, a LASFN9 (Berliner Glass GmbH, Germany, with refractive index at ) pressed on the waveguide, was turned by a computer-controlled motorized stage (Newport). The coupling angles were determined by detecting the light intensity at the end face of the waveguide using a biased photodiode (Siemens BPW 34 B) during a -angle scan. When the laser beam on the prism’s surface reflects back onto itself, the laser beam is perpendicular to the prism’s surface and equals zero. This was used as the reference angle. A scheme of the experimental setup can be found in Refs. 16, 25. The accuracy in measuring the synchronous angles using this method is about . This inaccuracy in the synchronous angle yields an inaccuracy of in the measurement of the effective refractive index, . A linearly polarized HeNe laser (JDS Uniphase, , ) was used for exciting the waveguide modes.
Fluorescence Emission Spectra and Loss Measurements
As previously found in studies of ion-exchanged waveguides,18, 19, 20, 21, 22, 23, 24 silver ions, molecules, clusters, and nanoparticles can cause optical electronic waveguide losses due to fluorescence excitation. When the evanescent field of the coupled light for a particular guided wavelength is designed to illuminate samples located on the waveguide (e.g., to carry out evanescent field fluorescence microscopy), additional fluorescence photons of a different wavelength are undesirable. However, they can be used as an internal intensity standard. Typically, coupling efficiencies to waveguides are not known without thorough investigation; therefore, an intrinsic light signal from the fluorescence of the waveguide can be useful for intensity calibrations.26
For optical loss measurements, polarized light of an argon ion laser (Melles Griot), with various laser lines from and maximum output powers from in both TE and TM polarizations, were coupled individually into the waveguides by a photoresist coupling grating.27 The intensity profile of the fluorescence light streak was viewed with an inverted microscope (Zeiss, Axiovert 25) and a digital camera (Diagnostic Instruments, Inc., Sterling Heights, Michigan) with frame-grabber software (Image Pro Express from Media Cybernetics, Bethesda, Maryland). Appropriate filters (dielectric long pass, Omega Optics) were used to filter the excitation wavelength propagating in the waveguide. The losses were calculated in dB/cm by analyzing the exponential decay of the emitted fluorescence light along the waveguide streak measured by means of the frame grabber software.
For obtaining the emission spectra of the excited Ag ions and agglomerates in the waveguide, fluorescence photons were collected along the streak by the microscope and spectrally analyzed by an adapted spectrometer via an optical fiber (Ocean Optics, Dunedin, Florida, HR-2000 spectrometer).
Determination of Penetration Depth of the Evanescent Field
The penetration depth of the evanescent intensity was quantified as the depth where the evanescent field dropped to 1/e of the intensity directly at the waveguide surface. The penetration depths of the different modes in the waveguide in both TE and TM polarizations were estimated from simulations of the field distribution, especially the evanescent field on top of the waveguide in an aqueous environment . The measured refractive index profile was implemented into the simulations. These simulations were performed by a commercially available software package called TRAMAX,28 which uses a transfer matrix formalism based on the Fresnel equations. The wavelength was chosen to be . To compare the evanescent fields directly and to be independent of experimental coupling conditions, all modes (TE modes) were normalized to their entire field integral along the three-layer waveguide structure.
HEK293 cells are a transformed human embryonic kidney cell line.29 Culture medium consisted of Dulbecco’s modified eagle medium (DMEM high glucose) supplemented with 10% heat-inactivated fetal bovine serum (FBS) and antibiotics (penicillin , streptomycin , amphotericin B ; Invitrogen, Burlington, Ontario). Waveguides were presoaked for at least in culture medium. Medium was then removed, and the cells were seeded directly on the waveguide. After incubation in 5% air at , samples were washed with phosphate-buffered saline (PBS) to remove any nonviable cells and loaded with fluorescent lipophilic probe 1, -dioctadecyl-3,3, -tetramethylindocarbocyanine perchlorate (DiI, Invitrogen) according to manufacturer’s instructions. The cells were imaged in phenol red-free DMEM (Invitrogen).
For TIRF microscopy, DiI-stained HEK293 cells were placed in Attofluor cell chambers (Molecular Probes, Eugene, Oregon) and positioned on the heated stage of a Zeiss inverted microscope (Axiovert 200M) equipped with a Zeiss TIRF slider. The filter set used (Zeiss 53) consisted of an excitation filter , a dichroic mirror FT 540, and an emission bandpass filter . Images were captured with a Zeiss Plan-Fluar oil-immersion objective lens at excitation wavelength of from a multiline argon ion laser . The images were recorded by a cooled CCD camera (AxioCam MR, Zeiss). With this microscope, epifluorescence and TIRF microscopy images can be taken of the same specimen by changing the angle of incidence on the coverslip. The optimal angle used for TIRF was approximately .
Waveguide Evanescent Field Fluorescence Microscopy
For imaging cell–substrate contact regions by waveguide evanescent field fluorescence microscopy, the stage of an inverted microscope (Zeiss, Axiovert 25) was modified to accommodate an xy-stage and a sample stage.11, 13, 14 A beam from a HeNe laser at a wavelength of and an output power of was coupled via a photoresist grating (grating constant ) into the ion-exchanged waveguide that was located on the sample stage of the microscope. The images were taken with a objective. A long-pass filter (3DR470LP, Omega Optics) was used for collecting the fluorescence emission from the DiI dye. A digital camera (Diagnostic Instruments, Inc., Sterling Heights, Michigan) with imaging software (Image Pro Express, Media Cybernetics, Bethesda, Maryland) was used for capturing the images on a computer.
Results and Discussion
All fabricated waveguides were characterized with -line spectroscopy. Using the measured coupling angles , the effective refractive indices , of both TE and TM modes were calculated.30 With increasing ion-exchange time at fixed temperature, the number of guided modes and the of a given mode increase, as previously reported.24 Figure 1 depicts the effect of ion-exchange time and temperature on for the modes of various multimode waveguides. It is evident that by increasing time or temperature, of a specific mode will increase. The results for higher order modes are similar (not shown).
We used the reversed Wentzel–Kramers–Brillouin (WKB)31 method for constructing the refractive index profile in the ion-exchanged glass waveguides from a set of identified . These profiles were then fitted successfully with a nonlinear diffusion profile proposed in Ref. 32, as demonstrated in Ref. 24;
Here, is the refractive index of the substrate, is the maximum increase in the refractive index at the surface, is the portion of exchanged ions, and are the diffusion coefficients of and , respectively, and is the effective diffusion depth. The effective diffusion depth, , is given by the relationshipis the diffusion time or ion-exchange time, and is the effective diffusion constant. The temperature dependence of below the glass transition temperature can be fitted to an Arrhenius equation:5, 33, 34 2, 3, 4, we find the time in minutes, and the temperature in Kelvin. and are constants.
Figure 2 depicts the linear behavior of the simulated versus for TE modes of various samples fabricated with pure silver nitrate. Table 1 summarizes the effective diffusion constant data , obtained from Eqs. 2, 3. With increasing temperature, the effective diffusion constant is increasing as depicted in Fig. 3 in the versus plot. With a linear fit, we determine and . By introducing the universal gas constant , the effective diffusion constant can be written in terms of an activation energy :4, 6 and using , the activation energy can be found as . The effective diffusion depth with a given temperature and time and the effective diffusion constant for a given temperature can be obtained by using Eqs. 5, 6 and constants and .
Measured surface refractive index change Δn and effective diffusion constant De as a function of exchange temperature T .
Table 1 also shows the surface refractive index change at the four different exchange temperatures. It can be seen that by increasing the temperature, increases, whereas it is known that is independent from the diffusion time.1, 3 By comparing the results for the diffusion coefficients of Ag ions in SG11 with those for Ag ions in BK7,35 Ag ions were found to have a much larger diffusion coefficient in SG11.
To use the waveguides for evanescent illumination purposes, precise knowledge about the evanescent field is needed—especially the penetration depth and the carried intensity in the illuminating field. The penetration depth of modes of a particular waveguide (for fabrication parameters of the waveguide, see legend) into the cover medium (here, water with a refractive index of ) increases with increasing mode number with a slight off-bend toward higher mode numbers and ranged between (Fig. 4 ). The influence of the silver concentration in the waveguide fabrication process on the intensity of the guided light in the evanescent field is shown in Fig. 5 . The normalized integrated intensity , calculated via the integrals of fields for the evanescent-guided fields, is plotted for two different four-mode waveguides as a function of the mode number. One waveguide was fabricated with pure , and the other by a mixture of 1% and 99% . The pure waveguide showed much higher integrated intensity in the evanescent field.
It is known that Ag ion exchange can lead to yellow-stained glasses due to absorption and photoluminescence (PL).18, 19, 20, 21, 22, 23, 24 Therefore, the fluorescence behavior of waveguides fabricated with pure silver salt and two different exchange times was investigated. Figure 6 shows fluorescence emission spectra taken at four excitation wavelengths ( , , , and ). The fluorescence emission spectra for the sample show contributions of two to three peaks, with a systematic highest peak at low wavelength (or high energy) and smaller peaks on the lower energy side [Fig. 6]. The sample shows less structured peaks but a systematic red shift in emission peak position with decreasing photon excitation energy [Fig. 6]. The coupling conditions into the waveguide mode were not controlled; therefore, the peak amplitudes cannot be compared. The effect of ion-exchange time on the formation of Ag compounds, aggregates, and possibly nanoparticles of various sizes and size distributions are the reason for this spectral behavior.18, 19, 20, 21, 22, 23
The losses of these waveguides were also investigated. Losses were found to depend on the fabrication procedure, the mode number, the polarization, and the wavelength propagating in the guide. Losses on the order of were observed, with loss values being typically larger for the higher number modes and smaller for lower number modes.
TIRF and Waveguide Evanescent Field Fluorescence Microscopy
To compare the waveguide evanescent field fluorescence microscopy with well-established TIRF microscopy, HEK293 cells were seeded on ion-exchanged waveguides and on TIRF coverslips and stained with DiI. Figure 7 shows a round and an elongated HEK293 cell. The left panels (a) and (b) are epifluorescence images, whereas the right panels (c) and (d) depict corresponding TIRF microscopy images. In the epifluorescence mode, the entire cell is depicted, showing high-intensity areas where dye accumulation occurs. In the TIRF mode, only focal regions of close contact between the cell membrane and the coverslip are visible. The different intensities here are due to a convolution of dye concentration and distance from the coverslip surface.
The waveguide evanescent field fluorescence image of similarly prepared HEK293 cells is depicted together with a bright-field image of these cells in Fig. 8 at a smaller magnification. In WEFF microscopy, a objective was used due to the lack of a objective. However, comparing images without investigating resolution issues is straightforward. The waveguide evanescent image reveals regional contacts of various cells with similar high contrast as in the TIRF images.
Images published so far obtained with waveguide evanescent field fluorescence microscopy often show—besides the evanescent image—contributions from epifluorescence. These contributions are produced by light scattered from the waveguides. This scattered light is able to excite fluorescent dyes outside the evanescent field in all parts of the cell that are stained. This leads to a weak but visible fluorescence emission of the entire cell’s plasma membrane. Therefore, the outlines of the cells can often be seen as weak, but additional, information in these mixed images. In Fig. 8 (WEFF), these areas of enhanced emission due to scattering are seen as the gray structures visible throughout the cell. This is in contrast to the crisp, mainly dot-like evanescent structures in the TIRF images [Fig. 7 and 7]. However, some scattering signal is evident in the TIRF images of Fig. 7 as well, leading to the grayish areas at locations where high dye concentration or cellular structures are detected in the epiflorescence mode.
The fine-tuning of the WEFF technology to achieve pure evanescent images or to achieve a controlled mixture of evanescent and scattering controlled epifluorescent images is still under development.36 In TIRF microscopy, two images can be taken at two different TIRF angles to yield a set of information—the entire cell and the focal regions of close contact. However, this takes time. In WEFF microscopy, on the other hand, images can be taken quickly with both sets of information in one image. This will be of advantage, when dynamic studies are performed and fast kinetic data are required.
Despite the losses and the autofluorescence of the waveguide, high-quality evanescent images could be obtained with the ion-exchanged SG11 waveguides showing the same quality with respect to contrast as the TIRF images.
Schott SG11 glass was used to fabricate waveguides as substrates for waveguide evanescent field fluorescence microscopy. It depicts all the necessary features of ion exchange but offers a much higher diffusion coefficient for Ag ions in comparison to standard glasses such as BK7 or soda-lime glasses. One can therefore fabricate waveguides as a microscopy substrate very economically at low temperatures and short exchange times. However, due to the high diffusion constant, these waveguide substrates cannot be autoclaved for disinfection. The heat will change the diffusion profile of the ions and therefore the optical performance. Therefore, chemical disinfection is required prior to seeding cells. Despite the autofluorescence that these waveguides show due to silver cluster formation and the losses associated with this effect, waveguide evanescent field fluorescence microscopy of HEK293 cells was successful. It compares very well with evanescent images taken from the same cells with a conventional TIRF microscope.
The authors thank Frank Thoma from Schott (Germany) for the glass samples and helpful discussions, the University of Western Ontario for various equipment grants, Natural Sciences and Engineering Research Council of Canada (NSERC), Canada Foundation for Innovation (CFI), Ontario Innovation Fund (OIF), Ontario Photonics Consortium (OPC) [through Ontario Research and Development Challenge Fund (ORDCF)], and the CRC Program of the Government of Canada for their kind financial support. M.N. and S.M. thank the University of Applied Sciences Wiesbaden for a travel grant.