Semiconductor lasers enable advances in nonlinear biophotonics

The successful demonstration of two-photon excitation fluorescence imaging of biotissues,¹ such as mouse kidney and brain cells, has spurred development of a variety of novel multiphoton imaging (MPI) technologies. For example, using near-IR optical pulses, MPI can peer deep into tissues for noninvasive profiling. Moreover, because the fluorescence occurs only at the focal point of IR optical pulses, very long-time imaging is possible. This minimizes the consumption of fluorescent material. Many MPI technologies depend on high peak-power (over a kilowatt) ultrashort optical pulse sources to efficiently induce various multiphoton, nonlinear optical effects inside biospecimens. But most current sources are mode-locked solid-state lasers, which are bulky, expensive, and require maintenance. Increasing the practical appeal of MPI will require ‘real-world’ alternatives.

Semiconductor laser diodes (LDs) have enjoyed substantial success as dependable, low-cost, high-performance light sources in information and communication technologies. In addition to their usefulness as mass-produced IT devices, LDs also have potential applications in scientific and technological measurement systems. Here we report evidence of their relevance to nonlinear bioimaging. In our research, LDs serve as key, easy-to-operate devices for generating stable picosecond optical pulses. A subsequent increase in optical power using low-noise and low-nonlinear-effect amplifiers raises the optical pulse peak power to more than a kilowatt, making it suitable for MPI. Because of the high peak power, efficient wavelength conversions using second-order nonlinear materials (e.g., ferroelectric optical crystals) and third-order materials (e.g., optical fibers, liquids) are also possible. Figure 1 presents the basic concept of our optical pulse source in schematic form.

Figure 1. Schematic for generating kilowatt peak power picosecond optical pulses using a semiconductor laser oscillator. LD: laser diode.

We have obtained optical pulses of less than 5ps duration by gain switching using a combination of a high-speed-response (>15GHz bandwidth) LD and an electric pulser that can generate 100ps bursts at flexible repetition frequencies.² This configuration may provide ultrashort optical pulse sources that are much simpler to set up and operate than mode-locked lasers. For nonlinear bioimaging, the peak power of optical pulses should be increased up to a few kilowatts. For this purpose, we have developed low-noise and low-nonlinear-effect erbium-doped optical fiber amplifiers. Using a 1550nm gain-switching LD as the pulse laser oscillator, we obtained 2kW peak-power optical pulses and second harmonic generation (SHG) (i.e., the light frequency is doubled) of over 50% conversion efficiency. We also applied this technique to two-photon fluorescence imaging (TPI) of the epithelial cells of a rat kangaroo kidney and convoluted tubules from a mouse kidney² (see Figure 2).

Figure 2. Two-photon fluorescence image of convoluted tubules in mouse kidney stained with Alexa Fluor 488 and excited by 1kW peak-power 774nm SHG optical pulses.

Figure 3. (a) Spectrum showing octave-wide supercontinuum light generated by the SHG output of high-peak-power 1.55μm optical pulses. (b) Prism-resolved photograph for the visible wavelength region of the supercontinuum light.

Note that although optical pulse compression to the femtosecond temporal region is not difficult, it is not necessary for TPI. Our present results clearly indicate that picosecond optical pulses are quite sufficient for practical use and obviate having to worry about pulse broadening due to optical dispersion for compressed femtosecond pulses. To extend the wavelength region of optical pulses for TPI with a variety of fluorescent materials, we generated ‘supercontinuum’ light by injecting SHG pulses into a photonic-crystal fiber (see Figure 3). After extracting the necessary wavelength component, we increased it using one of our optical fiber light-amplifying devices.^3–5

Mode-locked semiconductor laser diodes (MLLDs) are even more powerful optical pulse generators. We can easily synchronize a few picosecond bursts from different MLLD oscillators with low-jitter hybrid mode locking.⁶ This enables multicolor pulse excitation of biospecimens and can induce several different kinds of nonlinear effects that are useful for bioimaging, such as sum-frequency generation and Raman scattering. Starting from a gigahertz MLLD, the optical pulse repetition rate is decreased to subharmonic frequencies using a semiconductor optical amplifier (SOA) as the optical gating device. Pulse amplification to kilowatt peak power then becomes possible by employing optical fiber amplifiers as described for the gain-switching LD light source. We have adopted this configuration for 1550nm (and its SHG), 1030nm, and 980nm MLLDs, and have confirmed clear (high-resolution, high-contrast) TPI.^7,8 Figure 4 shows an example of TPI of mouse brain neurons expressing green fluorescent protein excited by kilowatt-peak-power amplified optical pulses from a 980nm MLLD.

Figure 4. Two-photon fluorescence image of mouse brain neurons expressing green fluorescent protein excited by amplified 980nm MLLD pulses.

Moreover, we recently demonstrated clear TPI using an all-semiconductor laser light source operating at a wavelength of 800nm, in which an MLLD oscillator and two-stage SOAs are employed to generate >100W-peak-power optical pulses.⁹ This result shows that such pulses at a variety of wavelengths can be obtained with many sets of MLLDs and SOAs, and that these light sources are very useful for nonlinear imaging.

We are currently also working on nonfluorescence (SHG, Raman effects) nonlinear-optic bioimaging technologies using our optical pulse sources. Increasing peak power to the megawatt level is an additional important step that would enable nanometer-scale material processing of biotissues (nanosurgery). These advanced laser technologies would also be beneficial for real-time local microtreatment of biospecimens in combination with nonlinear bioimaging methods.

The author would like to thank H. Tsubokawa, H.-C. Guo, J. Shikata, K. Sato, K. Takashima, K. Kashiwagi, N. Saito, H. Taniguchi, H. Ito, T. Hashimoto, Y. Kubota, M. Mure, Y. Iseki, H. Iizuka, and Y. Nakamura for their collaboration.