1. Introduction

Fluorescence labeling and detection is a powerful method in numerous applications, due to its high sensitivity and specificity [1–5]. Fluorescence emission is a spontaneous process and the emitted photons are distributed among all solid angles (4Pi). Therefore, fluorescence imaging is commonly performed with high numerical aperture (NA) optics, to achieve both a high spatial resolution and effective collection of emitted photons. To detect weak fluorescence signals from dark fluorophores, a highly sensitive detector with a large gain is frequently used. Stimulated emission (SE) was demonstrated as an advantageous method in detecting dark fluorophores [6]. The working principle of SE detection is to induce the electronic transition of an excited fluorophore into SE before its energy dissipates into other radiative or non-radiative decay processes. Therefore, the SE signals depend on both the population of excited state and the intensity of the stimulation beam. Without any saturation effects, the SE signal scales linearly with both excitation and stimulation beams, which renders its overall quadratic power dependence and the 3-D optical sectioning in imaging as a two-photon process. Additionally, spatial coherence of the stimulated fluorescence emission is maintained, leading to the emission in a narrow cone in the forward direction, similar to the case of second harmonic generation or other coherent optical processes. We have exploited the resulting spatial coherence for long working distance imaging of fluorescenctly labeled specimens [7,8].

The use of high NA optics for efficient collection of fluorescence would, however, limit the working distance and the depth of focus accordingly, which would, in turn, limit the size and the depth of the targeted objects. Separately, fluorescence lifetime microscopy is an effective tool to extract critical information on molecular dynamics, such as molecular conformation and the changes in the immediate nano-environment of a biological sample [9–12]. It is widely used in mapping pertinent cellular parameters and tissue characteristics, such as ion concentration [13], pH of the environment [14], and carious dental tissue identification [15]. Measuring fluorescence lifetime of targeted molecules is usually conducted by detecting the emission of spontaneously emitted photons, by methods of time- or frequency domains. In this work, we are demonstrating fluorescence lifetime measurement by the pump-probe scheme [16]. The temporal resolution of this method is typically determined by the pulse width of the lasers or the delay control between the pump-probe laser pair. We use electronic delay control and low NA optics to measure fluorescence lifetime of ATTO 647N dye and to achieve long distance detection. The fluorescence lifetime of ATTO 647N is approximately 3 ns, which would require an optical path length around a few meters to map out the delay. The relatively long length presents a substantial challenge in optical alignment and beam shaping if a mechanical translation stage is used. Electronics delay is used instead to overcome the above difficulties. Additionally, the dependences of SE on sample concentration and laser intensity are also characterized.

2. Materials and methods

The schematic of fluorescence lifetime detection with SE is shown in Fig. 1 . A pulsed diode laser (Picoquant, Berlin, Germany) that is operating at a wavelength of 635 nm acts as the excitation source and is synchronized with the mode-locked Ti:Sapphire laser (Mira900, Coherent, USA) through a fast photodiode (TDA 200, Picoquant, Berlin, Germany). The Ti:Sapphire laser is operated at 740 nm with a repetition rate of 76 MHz. The selected wavelengths closely match the absorption and emission spectra of the ATTO647N dye (ATTO-TEC, Germany), respectively.

 

Fig. 1 Schematic of the SE enabled long working distance fluorescence lifetime imaging setup. EOM: electro optical modulator. BP: band pass filter. DM: dichroic mirror. PD: photodiode.

Download Full Size | PDF

Depending on the fluorescence lifetime of the targeted fluorophores, various time delay schemes may be used. For very short fluorescence lifetime fluorophores (in the order of picoseconds or shorter), using precision mechanical translation stage for path length adjustment, as in the conventional pump-probe experiments, is a preferred choice, with the temporal resolution limited by the pulse width. For longer fluorescence lifetime species (in the order of nanoseconds or longer), electronics delay is preferred. The electronics delay can be generated by adjusting the length of the cable that connects the fast photodiode to the trigger of the pulsed diode laser driver, which would allow a delay step increment of approximately 0.12 ns with the temporal resolution limited by the timing jittering in electronics (~100 ps), instead of the pulse width [9]. Alternatively, high speed timing circuits can also be used for time delay control after adequate calibration. In this work, the cable length adjustment is adopted for its simplicity and stability.

The wavelength of the stimulation beam is chosen to match the red-shift region of the ATTO 647N’s emission spectrum to prevent its re-excitation. The excitation and stimulation beams were coupled collinearly through a dichroic mirror (720SP, Semrock) and directed into a commercially available galvono-mirror scanning unit (FV300, Olympus) for scanning imaging. A pair of low NA objectives (10X, NA 0.3) was used for focusing and collecting the transmitted stimulation signals, respectively. For longer distance imaging, a doublet lens with a focal length of 40 mm is used to focus the laser beam. The transmitted beam that carried the SE signal was detected using a biased Si detector (DET36A, Thorlab) with a large detection area to compensate for the beam movement as a result of galvo-mirror scanning. Additionally, the apertureless detection scheme also prevents the possible artifact attributed to the thermal lensing effect in pump-probe measurements [17]. Bandpass filters (747/33, Semrock) and a neutral density filter are used jointly to transmit the stimulation beam only, and to reduce the intensity of the stimulation beam to prevent saturation of the photodiode. To detect the SE signal, lock-in detection was used to recover the SE photons that were carried in the stimulation beam. To do so, the excitation beam was first modulated using an electro-optical modulator (EOM, ConOptics) at 30 KHz, which was driven by a function generator. The SE beam is then demodulated by a lock-in amplifier (SR830, Stanford Research USA) to recover the signal. The output of the lock-in amplifier was connected to the A/D converter of the galvono-mirror based laser scanning system (FV300) to reconstruct the image. For imaging, 256 by 256 pixels are acquired. Frame acquisition time is 10 min, which is currently limited by modulation frequency. To obtain time-resolved images the acquisitions are taken at various time delays with an interval of 0.5 ns. Since the SE signal is proportional to the population of excited molecules, the fluorescence decay curves can be obtained by correlating the image intensity as a function of time delay. The time constants of the lock-in amplifier are set to 1 ms and 300 μs for image acquisition and fluorescence lifetime measurement, respectively.

3. Results and discussion

The basic characterization of SE is shown in Fig. 2 , in which a dye concentration of 100 μM was used. The SE signal increases linearly with the excitation power, presented in Fig. 2(a), showing that the non-saturation condition was satisfied at the used excitation intensity (~10⁴ W/cm²). The non-saturation condition also prevented excessive photobleaching of the sample. The saturation intensity of organic dyes was approximately ~MW/cm², which was much higher than the excitation intensity that was used in this study. The ATTO 647N dye has been extensively used in stimulated emission depletion (STED) microscopy and exhibits excellent photo-stability under high intensity illumination. Figure 2(b) plots the dependence on the power of the stimulation beam. When the stimulation beam irradiates excited ATTO 647N molecules, SE or up-conversion occurs and contributes to the signal. Among the two processes, it has been shown that SE is the dominant fluorescence quenching process for ATTO 647N dyes [18]. The contribution from the up-conversion can thus be ignored. At low stimulation intensity, the signal exhibits linear dependence. When the power of the stimulation beam exceeds 30 mW (0.6 GW/cm²), SE signal begins to saturate due to the excited state depletion. The fluorescence depletion efficiency is defined as a function of the intensity of stimulation beam by η (r) = exp[-σI_ST(r)T_ST], where T_ST, σ and I_ST(r) denote the stimulation pulse duration, the cross section of the SE and the intensity of the stimulation pulse, respectively [19]. The loss of spontaneous fluorescence emission will be converted into stimulation photons. Thus, the SE signal as a function of intensity can be described by 1-η(r). The saturation condition observed herein has also been applied in diffraction-unlimited imaging, as demonstrated in saturation excitation microscopy [20]. The dye concentration dependence of SE signal is also shown in Fig. 2(c). The average power of excitation and stimulation are 100 μW and 20 mW, respectively. The condition of linear relationship allows quantitative analysis, as reported in stimulated Raman scattering (SRS) and anti-Stoke Raman scattering (CARS) microscopy [21, 22]. Using high NA objective, SE imaging has demonstrated sensitivity to a level of a few molecules [6]. In order to test the long distance detection capacity of our imaging system, a doublet lens with a focal length of 40 mm is used to focus the laser beam. The axial intensity response is shown in Fig. 2(d). The test sample was prepared by injecting 10 μM ATTO 647 N between two cover slips with 2 mm separation. Therefore, the axial intensity curve reveals the resulting convolution of point-spread function and the thickness of the sample. Notably, in the long-distance SE imaging setup, the image resolution is traded off for the extension of the working distance of the imaging system. As a comparison, spontaneously emitting fluorescence cannot be detected in the same setup, because the low NA optics is employed and the conventional biased Si detector used in the measurement provides insufficient sensitivity. For fluorescence detection, the high sensitivity detector such as PMT and APD is recommended and usually delicate to operate. Moreover, for detecting fluorescence in long distance, these detectors are inadequate due to high background noises including surrounding and stimulation beam.

 

Fig. 2 The dependences of SE signal. Power dependences on (a) the excitation and (b) stimulation beams. (c) The SE signal as a function of dye concentration, showing linear dependence. The concentration sensitivity is approximately a few μM. (d) Axial response of SE signal using a lens of 40 mm focal length.

Download Full Size | PDF

The SE signal as a function of pulse separation is shown in Fig. 3 . The gain of the stimulated beam is consistent with the quenching of spontaneous fluorescence. In the light quenching experiment, the intensity of fluorescence as a function of relative delay is described by Eq. (1),

 

Fig. 3 Fluorescence lifetime imaging. SE signal as a function of the relative time delay between the excitation and the stimulation pulses. The decay curve is obtained from a single scan position of the sample. Shown with the decaying curve are the SE images of ATTO 647 N prepared in Y-shape microfluidic channel. The image size is 600 μm × 600 μm.

Download Full Size | PDF

4. Conclusion

We have shown the feasibility of long working distance fluorescence lifetime imaging by utilizing electronic trigger control with pump-probe configuration. The scheme is simple, highly accurate, and cost-effective, which also prevents optical distortion by removing mechanical movement. Further reduction in size and cost can be easily achieved with all semiconductor light sources, while improvement the robustness simultaneously. By taking advantage of SE’s inherent spatial coherence, the long distance imaging capacity is also demonstrated. This capacity is projected to accommodate a broad range of application environment.