A novel concept for an optical parametric oscillator based on four-wave mixing (FOPO) in an optical fiber is presented. This setup has the ability of generating highly chirped signal and idler pulses with compressed pulse durations below 600 fs and pulse energies of up to 250 nJ. At a fixed pump wavelength of 1040 nm, the emerging signal and idler wavelengths can be easily tuned between 867 to 918 nm and 1200 to 1300 nm, respectively, only by altering the cavity length. With compressed peak powers >100 kW and a repetition rate of only 785 kHz, this source provides tunable intense ultra-short pulses at moderate average powers. This setup constitutes a stable, simple and in many ways superior alternative to bulk state-of-the-art OPO light converters for demanding biomedical applications and non-linear microspectroscopy.
© 2015 Optical Society of America
The optical fiber is a key component which has enabled the rapid advance of optical technologies in the field of communications, material processing and sensing applications. In spite of this, Titanium-Sapphire oscillators or optical parametric oscillators based laser systems, are due their ability to generate widely tunability nanojoule-level femtosecond pulses, the workhorse for non-linear molecular imaging. However, non-linear imaging techniques are now starting to make the transit to real-world applications and could therefore greatly benefit from a fiber-based solution. In this context, fiber-based frequency conversion based on four-wave mixing (FWM) has already enabled the development of compact and powerful laser sources delivering nearly ideal pulse parameters for coherent anti-Stokes Raman scattering microscopy . In contrast to the commonly used systems, these sources deliver pulse durations in the range of tens of picoseconds and are, therefore, unsuitable for applications where femtosecond pulse durations are beneficial, such as multi-photon excitation imaging. In this paper we address this issue and present a setup similar to the afore-mentioned laser source but able to deliver sub-picosecond pulse durations. Therefore, we emphasize on the generation of compressible energetic pulses in wavelength ranges beneficial for multi-photon imaging as an additional mode of operation for the setup presented in .
Some of the most popular fiber lasers are mode-locked soliton oscillators or oscillators with a dispersion map which can generate transform-limited ultra-short pulses at the output of the laser cavity. Depending on the parameters of the fibers, pulse energies in the range of nJ have been demonstrated with these approaches [2,3]. To generate higher pulse energies directly from the oscillator, the peak power of the circulating pulses has to be decreased in order to mitigate the impact of nonlinearities in the cavity. This can be done with normal dispersion oscillators, which generate linearly chirped pulses that can be externally compressed down to ultra-short pulse durations [4,5]. Here, the oscillating pulse stays positively chirped during each round-trip, keeping the accumulated non-linear phase (e.g. self-phase modulation, SPM) low even at higher pulse energies . During one round-trip the interplay of SPM induced spectral broadening and positive dispersion causes the chirped pulse to increase its spectral bandwidth and duration. If a spectral or temporal filter is inserted into these cavities, a stable steady-state solution can exist . The main drawback of these lasers is the limited accessible emission wavelength range given by the dopant. The accessible wavelength range of these systems is, therefore, usually limited to 1000-1100 nm  in the case of Yb based lasers, 1500-1650 nm in case of Er  and 1860-2100 nm in the case of Tm  and Ho  doped fibers.
Alternatively, fiber based frequency conversion techniques potentially allow the generation of tunable and energetic ultra-short pulses beyond the wavelength range limits that constrain rare earth doped fiber amplifiers. Thus, to generate pulses with wavelengths in between the spectral emission regions of rare-earth ions, fiber-based frequency conversion techniques such as super-continuum generation (SCG) , soliton self-frequency shift (SSFS)  and FWM [14,15] can be used. However, since SCG distributes the energy evenly across a part or the whole transparency window of the employed fiber host material, every wavelength is this window can be accessed, but it only stores a tiny fraction of the pump pulse energy. SSFS, on the other hand, operates in the anomalous dispersion regime and uses Raman scattering to shift part of the pump energy towards longer wavelengths. Thus, depending on the effective core area of the conversion fiber, only a limited amount of soliton energy is supported per soliton. As the core areas increase, higher energies can be allocated within the fundamental soliton , this strategy leads to the dispersion of the large core waveguide becoming relatively weak compared to the material dispersion. Consequently, in silica fibers energetic solitons can only be generated above 1250 nm . In this case exotic and expensive pump sources are needed to drive the SSFS. On the contrary, FWM offers a high degree of flexibility as the central wavelengths of the signal and idler are controlled by the phase matching condition which, in turn, is determined by the dispersion in the fiber. Thus, the central wavelengths of signal and idler can be engineered using special fiber designs, such as, for example, microstructured fibers [18–20]. In this work, we combine the benefits of the high pulse energies of normal dispersion lasers with the wavelength flexibility of FWM to generate a broadly tunable fiber based optical parametric oscillator with femtosecond pulse durations.
2. Four-wave mixing based optical parametric oscillator
The fiber optical parametric oscillator (FOPO) cavity (see Fig. 1) is seeded by an ytterbium-doped fiber MOPA system delivering 2.5 µJ pulse energy at a repetition rate of 0.78 MHz with a pulse duration of 60 ps. The MOPA is seeded by a narrowband mode-locked oscillator with a bandwidth of 40 pm at 1040 nm, which has been presented in detail elsewhere . Due to the relatively long pump pulse duration, amplification to the µJ energy level can be accomplished with only moderate amounts of non-linear pulse broadening even without the use of pulse stretchers and compressors.
The FOPO cavity consists of 11.5 cm photonic crystal fiber (PCF, one hole missing core design, hole diameter 1.55 µm, hole to hole distance 3.1 µm) for light conversion and amplification via four-wave-mixing (FWM), a polarization dependent output coupler, a 250 m piece of polarization maintaining single-mode step-index fiber and a variable delay line. The length of the cavity has been chosen to match the repetition rate of the pump laser [22,23]. Without closing the feedback-loop (i.e. 100% out-coupling) the gain bandwidth of the FWM process leads to broadband signal and idler radiation around 900 nm and 1300 nm, respectively. Since in this case the signal and idler are generated from noise, their spectral profiles roughly mimic the four-wave mixing gain characteristic (see Fig. 2) .
If part of the signal or idler light is coupled back through the delay line (to be synchronized with the repetition rate of the pump laser) a cavity is formed. The dispersion in the cavity causes the fed back pulses to be chirped and become longer than the incoming pump pulses. Therefore, only a small portion of the spectrum of the feed-back pulses will overlap with the pump pulses and amplified. If the gain provided by the four-wave mixing process is sufficient to overcome the losses in the cavity a laser signal will build up at the wavelength in resonance with the driving pump pulses. Figure 3(a) depicts the evolution of the idler spectrum with increasing pump energy. Since the cavity is highly dispersive, a small change of the cavity length shifts the resonant wavelength (by changing the spectral slice of the feed-back pulse that overlaps with the pump pulse), thus shifting the signal and idler central wavelengths . Therefore, the signal and idler wavelengths can be tuned within the gain bandwidth of the FWM fiber simply by varying the optical length of the delay line. The evolution of the idler spectra when the cavity length is changed in 1 cm steps is depicted in Fig. 3(b).
The feedback fiber provides normal dispersion to the oscillating pulses, which outweighs the anomalous dispersion of the microstructured fiber at the resonant idler wavelengths. Consequently, this cavity can be understood as a normal dispersion oscillator [5,6] with FWM acting as the gain mechanism. In contrast to conventional lasers of this type utilizing band pass filters the laser concept presented here employs dispersion filtering and temporal gain narrowing for spectral filtering. During each round-trip SPM and normal dispersion cause the resonant pulse to be broadened spectrally and stretched temporally such that only a limited spectral slice of the fed back FWM-signal temporally overlaps with the next pump pulse, a process called dispersion filtering . As the feed-back pulse is amplified mainly in the vicinity of the pump pulse peak, temporal gain narrowing acts as a spectral filter on the chirped feed-back pulse.
The broadest spectra (6.5 nm FWHM) were obtained with a feedback ratio (the part of the idler that is coupled back through the delay line and into the conversion fiber) of about 0.1% and with a coupled pump energy of 640 nJ at 1040 nm (see Fig. 4(a)). An average power of 24 mW at 1250 nm was measured at the output using dichroitic mirrors to isolate the idler radiation from the signal and the residual pump. This corresponds to a conversion efficiency of 4.7%. The pulses of the FWM driven normal dispersion cavity are linearly chirped and can be compressed by a grating pair. We used a reflective grating with 1250 lines per mm and placed them approximately 25 cm apart under the Littrow angle in double pass configuration. Using intensity autocorrelation we measured an autocorrelation duration of 800 fs. In order to determine the actual pulse duration, the recorded spectrum was Fourier-transformed. The residual phase after compression is predominantly due to third order dispersion originating from the feedback fiber, which is in good agreement with simulations (see next chapter). Consequently, in order to model the auto-correlation measurement third order dispersion has been added to the simulation of the Fourier-transform-limited pulse. Figure 4(b) shows the close match of the two curves which allows inferring that the actual pulse duration was 560 fs. This pulse duration, in turn, results in a pulse peak power of 21 kW with a compression efficiency of 40%. By employing compression gratings optimized for 1250 nm it should be possible to achieve 80% compression efficiency and >40 kW of peak power.
The pulse energy has been increased further by decreasing the conversion fiber length to 6 cm and increasing the pump pulse energy to 2.3 µJ. For stable conversion the feedback ratio has been reduced in order to correct for the additional spectral broadening in the feedback fiber by the larger non-linear phase. If the non-linear phase is not properly adjusted bandwidth and peak power are reduced. Additionally, the contrast between the pulse and the spectral background increased. At increased pump pulse energy we achieved an idler bandwidth of 4.7 nm. The pulses were compressible to an auto-correlation width of 1.6 ps. By taking into account the third order dispersion as described before a pulse duration of 960 fs was determined. The experimental results are depicted in Figs. 5(a) and 5(b).
For this configuration the conversion efficiency of pump to idler radiation has been enhanced to 10.9%, which corresponds to an extracted idler average power of 200 mW and a pulse energy of 250 nJ. After pulse compression with an efficiency of 40%, a peak power of 104 kW was obtained. Again, this value could be increased to >200kW with gratings optimized for this spectral range.
A numerical simulation has been conducted in order to gain a deeper understanding of the pulse dynamics within the cavity. We used the commercially available software Fiberdesk which solves the extended nonlinear Schrödinger equation by the split-step Fourier transform method. We implemented the elements of the cavity with a feedback ratio of , a pump energy of 250 nJ, 60 ps pump pulse duration, 11.5 cm of PCF fiber (single hole missing design, hole diameter 1.55 µm, hole to hole distance 3.1 µm) and 250 m of fused silica single cladding delay fiber with a 6 µm core diameter. Figure 6(a) shows the evolution of the output spectra of the FOPO during the first 100 round trips. With the chosen parameters, a steady state was reached after 30 round-trips. A bandwidth of 6.7 nm FWHM was obtained, which is in good agreement with the experimental result (see Fig. 4(a)). The temporal evolution of the pulse is displayed in Fig. 6(b). The pulse duration settles at 21 ps FWHM which corresponds to a reduction to 35% of the initial pump pulse duration of 60 ps. The average power conversion efficiency to the idler is 6.9% which matches well the measured conversion efficiency. The peak power conversion efficiency is 21.2% due to the shorter pulse durations of the signal and idler compared to the pump pulse. The pump energy required in the experiment could not be reproduced in the simulation (only ~50% of the pulse energy was needed in the simulation compared to the experiment) and is subject to further investigations.
Figure 6(c) depicts the spectrum of the idler pulses in the steady state. The spectral phase of the idler pulses has been plotted in the same graph revealing its parabolic shape. By compressing the pulses with grating pairs with the same line density as in the experiment (1250 lines per mm), the measured auto-correlation shape (red line in Fig. 6(d)) could be faithfully reproduced (blue line in Fig. 6(d)). This allows estimating the temporal profile of the compressed pulses (see black curve in Fig. 6(d)). The residual phase is dominated by 3rd order dispersion mostly coming from the delay fiber. Note that the obtained pulse duration of 560 fs is just ~18% longer than the transform limit of 475 fs. Lastly, Fig. 6(e) depicts the pulse evolution within a single round trip in steady state. In the feedback fiber the fed back pulse continuously lengthens by dispersion broadens by SPM. The FWM gain leads first to a narrowing of the spectrum. As the pulse is linearly chirped, the pulse duration decreased as well. As the gain saturates, the pulse is becoming broader again and the pulse length recovers.
This setup delivers pulse parameters which are currently unique for fiber-based frequency converters. In order to generate tuneable pulses with energies of 200 nJ from a standard bulk OPO cavity (usual repetition rates: ~80 MHz, longer OPO free space cavities are possible but are unrealistic regarding stability requirements) an output average power of already 16 W would be required. Considering inefficiencies in the conversion process, this would already require a highly sophisticated ultra-short pump system with average powers of ~100W. Thus, it can be argued that the presented setup surpasses current commercial bulk OPOs in terms of pulse energy. The key in the fiber based approach is the ultra long feedback cavity (>200 m) rendering low resonant repetition rates supporting high pulse energies at low average powers. The pump source required for the FOPO is therefore extremely simple as it only requires tens of picoseconds pulse durations, which can be directly amplified to microjoules of pulse energy in active fibers without the need of pulse stretching. Furthermore, as the FWM process is ideally pumped in the 1 µm range, a preconversion to the visible (as required for bulk OPO) is not required. The need for post-compression of the output pulses is, therefore, outweighed by the inherent stability of the fiber based source, its compactness, the simplicity of the pump source, the tuning method and the support for low repetition rates.
5. Application to three photon excited fluorescence (3PEF)
Deep tissue nonlinear microscopy requires intense diffraction limited laser radiation at moderate average powers. As it has been demonstrated elsewhere [26,27], the pump wavelength is highly important when it comes to delivering the applied energy relatively undistorted deep into tissue . For example, the absorption in human and animal tissue is determined mostly by the absorption of blood and water. Therefore, using pump wavelengths at around 1300 or 1700 nm leads to minimal absorption and allows scanning up to 1.6 mm in depth without causing photo-induced damage. Additionally, it has been shown that if the degree of non-linearity of the signal generation is higher than the degree of non-linearity of the process that causes photo-toxicity, low repetition rate excitation generates higher signal powers or, alternatively, allows obtaining deeper penetrations . In the case of three photon excited auto-fluorescence (3PEF), the signal intensity is proportional to the intensity of the pump radiation to the 3rd power . In this parameter range the dominant effect leading to tissue damage is two photon absorption [31,32]. Hence, the photo-damage increases proportional to the square of the intensity. Since THG and 3PEF are proportional to the third power of the intensity, it is possible to simultaneously optimize the THG and 3PEF signals and reducing the photo-induced damage by increasing the peak intensity and reducing the laser repetition rate. Thus, using these estimations, Fig. 7(a) shows the expected normalized 3PEF signal generated at different repetition rates for constant pump pulse durations . The pulse peak power for each data point was adjusted to cause the same amount of phototoxicity induced by two photon damage effects according to the following equation . As can be seen, lower repetition rates lead to higher 3PEF signal powers. However, the lowest value of the repetition rate is limited by the desired pixel acquisition time.
In the experiment conducted in  up to 1.5 nJ with a pulse duration of 140 fs at 1280 nm and a repetition rate of 80 MHz was used to scan up to 1.6 mm deep into a mouse cortex. Assuming a two-photon absorption photo-damage mechanism as discussed above, in principle up to 27 nJ of pulse energy can be used for ~1 ps pulses at the same level of photo-damage, since higher pulse peak powers can be used at 0.78 MHz than at 80 MHz pulse repetition rate. This leads to a theoretical signal increase of a factor of 3.8 compared to the laser source in .
In order to acquire the image Fig. 7(b) we used a customized laser scanning microscope equipped with a 0.40 NA objective. The signals were collected in forward detection. As a proof-of-principle experiment we used a sample without any fluorescent markers, unlike in . Figure 7(b) depicts a section of a rabbit aorta combining 3PF (green), third harmonic generation (THG, blue) and second harmonic generation (SHG, red) images. The laser was blocked using a short pass filter at 750 nm (Semrock). In addition, the 3PEF signal was filtered by a bandpass filter . For recording the THG signal a bandpass filter centered at 415 nm has been used . The second harmonic generation signal at 625 nm was isolated with a bandpass filter . Even without using a fluorescent label, a strong signal coming from the (much weaker) auto-fluorescence process could be detected. As discussed above, this laser source can potentially provide even higher signal powers than the laser system used in , if the sample is similarly labelled using high intensity fluorescent markers. Therefore, we believe that this laser source allows imaging tissue layers buried up to 1 mm deep below the surface as demonstrated in  when utilizing an appropriate sample and imaging setup.
In future works we want to extend the tunable range by employing a continuously wavelength tunable pump laser. We expect to be able to generate signal wavelengths between 800 and 970 nm and idler wavelengths between 1150 and 1400 nm.
In conclusion, we have presented the first low repetition rate femtosecond FWM-based normal dispersion fiber OPO. The setup is capable of generating energetic picosecond pulses in the wavelength ranges of 867 to 918 nm and between 1200 and 1300 nm. Continuous tuning, exploiting dispersion tuning, is performed solely by changing the cavity length of the FOPO cavity. The external compression of the pulses yields clean ultra-short pulses of only 560 fs and 21 kW of peak power. By decreasing the feedback ratio in the cavity, pulses with durations of 960 fs and energies of 250 nJ can be generated and externally compressed to achieve 104 kW of peak power. By employing fibers, our setup supports low pulse repetition rates with a compact and stable cavity which ultimately decouples high peak powers from high average powers making this laser setup the ideal source for nonlinear spectroscopy and imaging at moderate average powers and phototoxic parameter levels. This source combines the versatility of four-wave mixing in photonic crystal fibers and the outstanding performance and reliability of normal-dispersion mode-locked fiber lasers. This laser source enables the broadening of the area of operation for biomedical imaging and material studies, by offering tunable energetic femtosecond pulses from a reliable and robust fiber based system. This development ushers a new era in which optical parametric light converters are becoming as simple and reliable light sources for modern life science.
References and links
1. T. Gottschall, T. Meyer, M. Baumgartl, B. Dietzek, J. Popp, J. Limpert, and A. Tünnermann, “Fiber-based optical parametric oscillator for high resolution coherent anti-Stokes Raman scattering (CARS) microscopy,” Opt. Express 22(18), 21921–21928 (2014). [CrossRef] [PubMed]
4. W. H. Renninger, A. Chong, and F. W. Wise, “Pulse shaping and evolution in normal-dispersion mode-locked fiber lasers,” IEEE J. Sel. Top. Quantum Electron. 18(1), 389–398 (2012). [CrossRef] [PubMed]
7. C. Aguergaray, R. Hawker, A. F. J. Runge, M. Erkintalo, and N. G. R. Broderick, “120 fs, 4.2 nJ pulses from an all-normal-dispersion, polarization-maintaining, fiber laser,” Appl. Phys. Lett. 103(12), 121111 (2013). [CrossRef]
8. O. G. Okhotnikov, L. Gomes, N. Xiang, T. Jouhti, and A. B. Grudinin, “Mode-locked ytterbium fiber laser tunable in the 980-1070-nm spectral range,” Opt. Lett. 28(17), 1522–1524 (2003). [CrossRef] [PubMed]
9. M. R. Mokhtar, C. S. Goh, S. A. Butler, S. Y. Set, K. Kikuchi, D. J. Richardson, and M. Ibsen, “Fibre Bragg grating compression-tuned over 110nm,” Electron. Lett. 39(6), 509–511 (2003). [CrossRef]
10. W. A. Clarkson, N. P. Barnes, P. W. Turner, J. Nilsson, and D. C. Hanna, “High-power cladding-pumped Tm-doped silica fiber laser with wavelength tuning from 1860 to 2090 nm,” Opt. Lett. 27(22), 1989–1991 (2002). [CrossRef] [PubMed]
11. A. S. Kurkov, E. M. Sholokhov, O. I. Medvedkov, V. V. Dvoyrin, Y. N. Pyrkov, V. B. Tsvetkov, A. V. Marakulin, and L. A. Minashina, “Holmium fiber laser based on the heavily doped active fiber,” Laser Phys. Lett. 6(9), 661–664 (2009). [CrossRef]
12. S. L. Alfano and S. L. Shapiro, “Emission in the region 4000 to 7000 a via four-photon coupling in glass,” Phys. Rev. Lett. 24(11), 584–587 (1970). [CrossRef]
14. D. Nodop, C. Jauregui, D. Schimpf, J. Limpert, and A. Tünnermann, “Efficient high-power generation of visible and mid-infrared light by degenerate four-wave-mixing in a large-mode-area photonic-crystal fiber,” Opt. Lett. 34(22), 3499–3501 (2009). [CrossRef] [PubMed]
15. W. Wadsworth, N. Joly, J. Knight, T. Birks, F. Biancalana, and P. Russell, “Supercontinuum and four-wave mixing with Q-switched pulses in endlessly single-mode photonic crystal fibres,” Opt. Express 12(2), 299–309 (2004). [CrossRef] [PubMed]
16. K. Wang, N. Horton, K. Charan, and C. Xu, “Advanced fiber soliton sources for nonlinear deep tissue imaging in biophotonics,” IEEE J. Sel. Top. Quantum Electron. 20, 6800311 (2014).
17. M. Chan, S. Chia, T. Liu, T. Tsai, M. Ho, A. A. Ivanov, A. M. Zheltikov, J. Liu, H. Liu, and C. Sun, “1.2- to 2.2-µm tunable raman soliton source based on a Cr :forsterite laser and a photonic-crystal fiber,” IEEE Photonics Technol. Lett. 20(11), 900–902 (2008). [CrossRef]
18. J. E. Sharping, M. Fiorentino, P. Kumar, and R. S. Windeler, “Optical parametric oscillator based on four-wave mixing in microstructure fiber,” Opt. Lett. 27(19), 1675–1677 (2002). [CrossRef] [PubMed]
19. J. D. Harvey, R. Leonhardt, S. Coen, G. K. L. Wong, J. C. Knight, W. J. Wadsworth, and P. S. J. Russell, “Scalar modulation instability in the normal dispersion regime by use of a photonic crystal fiber,” Opt. Lett. 28(22), 2225–2227 (2003). [CrossRef] [PubMed]
20. J. E. Sharping, “Microstructure fiber based optical parametric oscillators,” J. Lightwave Technol. 26(14), 2184–2191 (2008). [CrossRef]
21. T. Gottschall, M. Baumgartl, A. Sagnier, J. Rothhardt, C. Jauregui, J. Limpert, and A. Tünnermann, “Fiber-based source for multiplex-CARS microscopy based on degenerate four-wave mixing,” Opt. Express 20(11), 12004–12013 (2012). [CrossRef] [PubMed]
22. F. Kienle, P. S. Teh, S.-U. Alam, C. B. E. Gawith, D. C. Hanna, D. J. Richardson, and D. P. Shepherd, “Compact, high-pulse-energy, picosecond optical parametric oscillator,” Opt. Lett. 35(21), 3580–3582 (2010). [CrossRef] [PubMed]
23. T. Südmeyer, J. Aus der Au, R. Paschotta, U. Keller, P. G. R. Smith, G. W. Ross, and D. C. Hanna, “Femtosecond fiber-feedback optical parametric oscillator,” Opt. Lett. 26(5), 304–306 (2001). [CrossRef] [PubMed]
24. M. Baumgartl, T. Gottschall, J. Abreu-Afonso, A. Díez, T. Meyer, B. Dietzek, M. Rothhardt, J. Popp, J. Limpert, and A. Tünnermann, “Alignment-free, all-spliced fiber laser source for CARS microscopy based on four-wave-mixing,” Opt. Express 20(19), 21010–21018 (2012). [CrossRef] [PubMed]
26. N. G. Horton, K. Wang, D. Kobat, C. G. Clark, F. W. Wise, C. B. Schaffer, and C. Xu, “In vivo three-photon microscopy of subcortical structures within an intact mouse brain,” Nat. Photonics 7(3), 205–209 (2013). [CrossRef] [PubMed]
28. D. Kobat, M. E. Durst, N. Nishimura, A. W. Wong, C. B. Schaffer, and C. Xu, “Deep tissue multiphoton microscopy using longer wavelength excitation,” Opt. Express 17(16), 13354–13364 (2009). [CrossRef] [PubMed]
29. P. Theer, M. T. Hasan, and W. Denk, “Two-photon imaging to a depth of 1000 µm in living brains by use of a Ti:Al2O3 regenerative amplifier,” Opt. Lett. 28(12), 1022–1024 (2003). [CrossRef] [PubMed]
30. S. W. Hell, K. Bahlmann, M. Schrader, A. Soini, H. M. Malak, I. Gryczynski, and J. R. Lakowicz, “Three-photon excitation in fluorescence microscopy,” J. Biomed. Opt. 1(1), 71–74 (1996). [CrossRef] [PubMed]
32. K. König, P. T. C. So, W. W. Mantulin, and E. Gratton, “Cellular response to near-infrared femtosecond laser pulses in two-photon microscopes,” Opt. Lett. 22(2), 135–136 (1997). [CrossRef] [PubMed]