1. Introduction

Two-photon fluorescence microscopy (TPFM) has become a powerful tool for biological imaging [1, 2]. TPFM systems usually employ solid-state mode-locked pulsed lasers (MLLs) [3, 4]. It is expected that the development of highly practical pulse sources featuring compactness, ease of handling, excellent stability, and low cost will encourage the further widespread acceptance of TPFM.

Yokoyama et al. [5] have developed a picosecond pulse source with an emission wavelength of 780 nm by using a combination of a high-power picosecond pulse source based on an electrically controlled gain-switched laser diode (GS-LD) and a second-harmonic generator. They demonstrated two-photon fluorescence imaging of a biological sample using this pulse source with a laser-scanning microscope (LSM).

A pulse source with an emission wavelength in the range 700 to 1000 nm is usually used in TPFM in order to observe two-photon-excited visible fluorescence. A pulse source with an emission wavelength in the range 900 to 1000 nm is especially important, because the most popular fluorescent proteins, such as green fluorescent protein (GFP), yellow fluorescent protein (YFP), and red fluorescent protein (RFP), have two-photon excitation peaks in that wavelength range [2].

In this study, we developed a GS-LD based pulse source with an emission wavelength of 980 nm, and using the developed pulse source in combination with an LSM, we also successfully demonstrated two-photon fluorescence imaging of biological samples, namely, Alexa488-stained Glomeruli and convoluted tubules of a mouse kidney section.

2. Pulse source setup

The configuration of the optical pulse source is shown in Fig. 1. The pulse source consisted of a GS-LD driven by an electrical pulse generator, a pulse compressor, a pulse reshaper, and optical amplifiers. All components of the pulse source were connected to each other by optical fibers. The optical pulse generated by the GS-LD had a large red chirp, originating from the carrier dynamics of the LD. The temporal width of the optical pulse was compressed by applying positive group-velocity dispersion (GVD) to compensate for the red chirp, reducing the optical pulse width to about a few picoseconds [6]. The pulse had a pedestal, however, which was reduced by using an optical band-pass filter (BPF) to reshape the pulse. Finally, optical amplifiers boosted the optical power of the pulse train to achieve sufficient power for TPFM.

2.1 GS-LD

A single-mode vertical surface emitting laser (VCSEL; U-L-M Photonics GmbH, Ulm, Germany) operating at 980 nm and having a threshold current of 0.4 mA and a side-mode suppression ratio of 40 dB was gain-switched in this experiment. Electrical pulses having a peak-to-peak voltage of 2.8 V, a pulse width of 250 ps, and a repetition rate, f_rep, of 500 MHz were applied to the VCSEL. The resulting optical pulse emitted from the GS-LD had a pulse width of 31 ps, a center wavelength of 980 nm, an optical spectral width of 0.5 nm, and a pulse energy of 0.20 pJ. The pulse width was measured using a second harmonic generation (SHG) intensity auto-correlation.

2.2 Pre-amplification and pulse reshaping

The optical pulse was pre-amplified by a commercially available semiconductor optical amplifier (SOA1; Dora Texas Corp., Pearland, USA), which had a small-signal gain of 23 dB, a gain bandwidth of 20 nm, and a saturation power of about 5 mW. Isolators were placed before and after the SOA to avoid lasing of the SOA and optical feedback to the VCSEL.

A dielectric thin film multilayer optical BPF, having a 3-dB bandwidth of 0.6 nm, placed after the SOA1 suppressed amplified spontaneous emission (ASE) noise of the SOA1. Since the tail of the pulse in the time domain can also be trimmed by filtering [5], the center wavelength of the BPF was set at a slightly shorter wavelength than the center wavelength of the optical pulse generated from the GS-LD by tuning the incident angle at the BPF.

2.3 Pulse compression

The filtered pulse, which had a pulse width of 21 ps, still had a large red chirp, which was compensated by a 600-m-long standard silica-glass single-mode fiber having a positive GVD value of 14.4 ps² and a propagation loss of 1.2 dB at 980 nm. The compressed optical pulse had a width of 3.5 ps, which was estimated by the SHG intensity auto-correlation measurement.

2.4 Post-amplification

The compressed pulse was amplified up to a pulse energy of 10 pJ by another semiconductor optical amplifier (SOA2) which had the same specifications as the SOA1. The SOA2 was controlled on/off to be sub-harmonically synchronized with the repetition frequency f_rep of the incident optical pulse train in order to reduce the ASE noise generated during the pulse interval. When inactive (off), the SOA2 acted as an absorber instead of an amplifier. Therefore, the on/off driving of the SOA2 was highly effective in avoiding excess ASE noise, when the amplified pulse interval was longer than the carrier recovery time of the SOA2. In this experiment, the SOA2 was used as both a pulse picker and a post-amplifier. The repetition frequency f_rep of the SOA2 was set at 1, 2, 5, 10, and 20 MHz, which are sub-harmonic of the f_rep of the incident pulse train [7].

The output pulse from the SOA2 was introduced to an Yb-doped fiber amplifier (YDFA) module which we developed. In order to produce optical gain at 980 nm with the YDFA, we used a pump laser diode with an optical power of 150 mW and an emission wavelength of 907 nm. The YDFA had a small-signal gain of 30 dB, a gain bandwidth of 4 nm, and a saturation power of about 35 mW. The length of the fiber in the YDFA was short and the YDFA had a spatial output, instead of a fiber output, to reduce the influence of the self-phase modulation effect [5, 7]. Isolators were placed at the input and output ends of the YDFA module to prevent lasing.

 

Fig. 1. Pulse source setup. VCSEL: vertical cavity surface emitting laser, SOA: semiconductor optical amplifier, BPF: band-pass filter, YDFA: Yb-doped fiber amplifier.

Download Full Size | PDF

3. Experimental Results

3.1 Temporal and spectral traces

The intensity auto-correlation and the spectral trace of the pulse source output are shown in Fig. 2(a) and 2(b). The full width at half maximum (FWHM) of the autocorrelation was 5.4 ps, and the FWHM of the pulse was estimated to be 3.5 ps, assuming that the pulse had a sech² temporal shape. The spectral width was 0.6 nm, and the time-bandwidth product of the pulse was 0.66. The pedestal accompanying the pulse was negligibly small. When f_rep was smaller than 10 MHz, the peak power of the pulse was higher than 1 kW. In Fig. 2(b), small ripples appear on the spectrum due to the residual reflection from the two anti-reflection (AR) coated facets of the SOAs.

 

Fig. 2. Intensity auto-correlation trace (a) and optical spectrum (b).

Download Full Size | PDF

3.2 Two-photon fluorescence imaging

Two-photon fluorescence imaging was demonstrated with the setup shown in Fig. 3 [5]. The TPFM system consisted of the developed pulse source, an X-Y beam scanning unit (Olympus, FV300), and a microscope (Olympus, IX71). The collimated output beam having a diameter of 2.8 mm of the pulse source was coupled into the X-Y beam scanning unit through two steering mirrors. The numerical aperture (NA) of the objective lens (Olympus, UPlanSApo 60xW UIS2) used in this experiment was 1.20 with water immersion, and the two-photon fluorescence was detected by a photomultiplier tube (PMT; Olympus, FV5-PMTI2).

 

Fig. 3. Two-photon fluorescence imaging setup. PMT: photomultiplier tube. NA of the objective lens was 1.2 with water immersion.

Download Full Size | PDF

Two-photon fluorescence imaging of a mouse kidney section (Invitrogen, Molecular Probes^TM F24630) prepared on a slide was successfully demonstrated with the developed 980-nm GS-LD based pulse source; an image is shown in Fig. 4. The average power incident on the X-Y beam scanning unit was 35 mW and f_rep was 10 MHz. The average power after the objective lens was estimated at about 2 mW with a separately characterized transmittance data of the objective lens and a power before the objective lens. AlexaFluor488 wheat germ agglutinin, a green-fluorescent lectin, was used to label elements of the glomeruli and convoluted tubules. The filamentous actin prevalent in glomeruli and the brush border were stained with red-fluorescent AlexaFluor568 phalloidin, and the nuclei were counterstained with the blue-fluorescent DNA stain DAPI. Although the sample was stained with three kinds of fluorescent probes, only AlexaFluor488 emitted the two-photon fluorescence, which was confirmed by comparison with an image obtained by a conventional confocal LSM.

 

Fig. 4. Two-photon fluorescence imaging of a mouse kidney stained with AlexaFluor488.

Download Full Size | PDF

4. Concluding remarks

We have developed a practical pulse source with an emission wavelength of 980 nm consisting of a GS-LD, an optical-fiber-based pulse compressor, a BPF-based pulse reshaper, and optical amplifiers. We have also successfully demonstrated two-photon fluorescence imaging of biological samples using the developed pulse source in combination with an LSM. The optical pulse generated by the pulse source had a pulse width of 3.5 ps and kilowatt-level peak power. The GS-LD based pulse source is highly practical in terms of its environmental stability, ease of handling, compactness, and low cost. The successful demonstration of two-photon fluorescence imaging confirmed the potential of the developed pulse source for TPFM applications. It is expected that TPFM employing the developed pulse source will be applied to various kinds of imaging of biological samples marked with fluorescent dyes and proteins, such as GFP, YFP, and RFP.