Expand this Topic clickable element to expand a topic
Skip to content
Optica Publishing Group

Four-wave mixing seeded by a rapid wavelength-sweeping FDML laser for nonlinear imaging at 900 nm and 1300 nm

Open Access Open Access

Abstract

Four-wave mixing (FWM) enables the generation and amplification of light in spectral regions where suitable fiber gain media are unavailable. The 1300 nm and 900 nm regions are of especially high interest for time-encoded (TICO) stimulated Raman scattering microscopy and spectro-temporal laser imaging by diffracted excitation (SLIDE) two-photon microscopy. We present a new, to the best of our knowledge, FWM setup where we shift the power of a home-built fully fiber-based master oscillator power amplifier (MOPA) at 1064 nm to the 1300-nm region of a rapidly wavelength-sweeping Fourier domain mode-locked (FDML) laser in a photonic crystal fiber (PCF) creating pulses in the 900-nm region. The resulting 900-nm light can be wavelength swept over 54 nm and has up to 2.5 kW (0.2 µJ) peak power and a narrow instantaneous spectral linewidth of 70 pm. The arbitrary pulse patterns of the MOPA and the fast wavelength tuning of the FDML laser (419 kHz) allow it to rapidly tune the FWM light enabling new and faster TICO-Raman microscopy, SLIDE imaging, and other applications.

© 2023 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement

Faster imaging is useful in many applications and allows for imaging without motion blur or being able to measure cells in rapid succession. Rather slow but promising imaging modalities are two-photon and Raman microscopy, which are widely used in medicine and biology [1]. A fast version of two-photon microscopy is spectro-temporal laser imaging by diffracted excitation (SLIDE), which uses a fast-sweeping amplified Fourier domain mode-locked (FDML) laser and a grating to improve the beam-scanning speed on the sample [2]. The fast wavelength changing of the FDML laser results in the high line rate of SLIDE. This can be used, for example, in flow cytometry. At 900 nm, dyes based on the green fluorescent protein can be excited with two-photon microscopy [3] and used for brain activity measurements [4]. Not only does SLIDE microscopy benefit from an amplified FDML laser, but also so-called time-encoded (TICO) stimulated Raman microscopy, where the fast wavelength tuning is used to rapidly measure different Raman bands [5]. TICO-Raman is a promising technology, however currently it suffers from the low power levels of the FDML probe laser. Besides the fast wavelength tuning, a high-power probe laser could allow for faster TICO-Raman microscopy and a detection in epi-direction for future use in Raman endoscopy. In the future, we want to combine TICO-Raman with optical coherence tomography (OCT) and only probe the Raman signal at specific locations. Usually titanium-sapphire (Ti:Sa) lasers in combination with optical parametric oscillators are used for stimulated Raman microscopy [3,6,7], which are bulky, cost intensive, and require pulse prechirping if used in endoscopes. Alternatively fiber lasers can bypass all these problems [8]. Unfortunately, no appropriate fiber gain media are available in the 900-nm and 1300-nm wavelength regions. Recently, four-wave mixing (FWM) in a photonic crystal fiber (PCF) was presented using a 1040-nm fiber laser as pump and a 1330-nm tunable ring laser as a seed laser to slowly tune the seed wavelength for Raman sensing [9]. Based on this concept, an even faster wavelength-sweeping version can be achieved if an FDML laser is used as a seed laser. Additionally, other research groups have utilized FWM, even in the absence of a tunable seed laser, to generate wavelengths where fiber gain media are not readily available [10,11].

Here, we present FWM using a home-built 1064-nm master oscillator power amplifier (MOPA) fiber laser as pump and an FDML laser with a center wavelength at 1300 nm as seed. FDML lasers are rapidly and periodically wavelength-sweeping at up to 3 MHz (here 419 kHz); they have a narrow instantaneous linewidth and low laser noise. They exhibit this continuous wave (cw) sweep operation up to 170 nm [1214]. Here, by amplifying an FDML idler laser in a PCF through FWM, we generate spectral narrowband signal pulses in the 900-nm region. These 900-nm, 80-ps-long pulses have a peak power of up to 2.5 kW (0.2 µJ) and an instantaneous spectral linewidth of 70 pm over an adjustable span of 54 nm. The wavelength of the signal pulses is determined by synchronizing the 1064-nm pump pulse with the timing of the FDML sweep. As the 1064-nm pump laser can be triggered freely with “pulse-on-demand” capability, the amplified wavelength is determined by the temporal overlap between the pump pulse and the current wavelength of the cw FDML wavelength sweep. Here, our FDML laser is sweeping over 105 nm at 419 kHz which determines the maximum repetition rate of pulses with the same wavelength. The initial pulses of the 1064-nm MOPA laser are generated by an electronically modulated laser diode, whose repetition rate, pulse timing, and pulse length can be chosen freely before entering the amplifiers.

The setup consists of four parts (Fig. 1): the 1064-nm MOPA laser, the 1300-nm FDML laser, the free space combining of both lasers, and the FWM in the PCF with subsequent filtering to select either pump, signal, or idler. All four parts are explained in detail in the following.

 figure: Fig. 1.

Fig. 1. FWM setup. Green: 1064-nm MOPA laser used as pump. EOM, electro-optic modulator; YDFA, ytterbium-doped fiber amplifier. Red: free space combining of the 1064-nm MOPA laser and the 1300-nm FDML laser. BP, bandpass filter; LP, longpass filter; SP, shortpass filter. Blue: FWM in the PCF with filters afterwards.

Download Full Size | PDF

To achieve the necessary high peak power of the 1064-nm MOPA laser, a five-stage amplifier increases the power of an electronically modulated laser diode [15,16]. There are three preamplifiers and two large core double-cladding fiber amplifiers to increase the peak power up to 100 kW. For a narrow spectral full width at half maximum (FWHM) of the four-wave mixed light, spectrally narrow pump and seed lasers are needed [17,18]. The amplification in each stage was adjusted to have a spectrally narrowband output. The 1064-nm laser diode (CMDFB1064A, II-VI Laser Enterprise, USA) is premodulated to emit a 20-ns-long pulse to reduce leaking light through the following electro-optic modulator (EOM), which we modulate here for 100-ps pulses. The electronic control signals for the modulation of the diode and the EOM are generated by an arbitrary waveform generator enabling pulse-on-demand capability. Additionally, the pulse repetition rate can be freely changed between 100 kHz and 1 GHz and the pulse length between 30 ps and 10 ns. The three core pumped stages use a 6-µm core polarization-maintaining (PM) fiber (PM-YSF-HI-HP, Nufern/Coherent, USA). The amplification in these stages is optimized for maximal peak power with minimal amplified spontaneous emission (ASE). ASE is more problematic at lower repetition rates (<500 kHz) and needs a lower amplification per stage. In the two DC stages, a larger core size is used, since nonlinear effects in the fiber take place and result in a broadening of the spectrum, which is not desired for our applications. Self-phase modulation (SPM) and Raman scattering are the main problems in our setup at high intensities in the optical fiber. Hence, the main amplification takes place in the last stage to shorten the interaction length of the high-power pulses. In the first DC stage, the core size is 10 µm (Yb1200-10/125DC-PM, nLight, USA). We achieve up to 4 kW peak power maintaining a small linewidth. The second DC stage uses a bigger large mode area (LMA) fiber (Yb1200-25/250PC-PM, nLight). With this 25-µm core fiber we achieved 100 kW of peak power with low but measurable spectral broadening due to SPM. To avoid potential damage, we did not evaluate the maximum achievable power. Also, for the setup here we only used 30 kW of peak power due to saturation of the FWM process.

To seed the FWM process, an FDML laser is used. FDML lasers are fast wavelength-sweeping cw fiber lasers, mostly used for OCT [12]. An intra cavity home-built tunable fiber Fabry–Perot (FFP) filter was swept over ∼105 nm at 419 kHz and we previously measured the “instantaneous linewidth” to be as narrow as 12 kHz [19]. The length of the laser cavity is precisely designed to match the round trip time of the light with the inverse filter drive frequency. This leads to a stationary operation with the FDML laser emitting repetitive cw wavelength sweeps with an average sweeping velocity of 88 nm/µs. The 1300-nm FDML laser is set up in a sigma ring configuration (Fig. 2). The main parts of the FDML laser are the semiconductor optical amplifier (BOA1130S, Thorlabs, USA), the fiber delay spool, and the FFP [20]. To have a stable FWM process and therefore stable FWM pulses, the output noise of the FDML needs to be minimized. A chirped fiber Bragg grating (cFBG) (TeraXion, Canada) in combination with different fiber types (Hi1060, SMF28e+, LEAF) in the fiber spool, is used for highly precise dispersion compensation, which leads to a long coherence length and a very low noise output of the FDML [13]. To keep the FDML laser running in the so-called “sweet spot”, where the noise decreases almost down to the shot noise limit (at several mW output power and ∼50 GHz measurement bandwidth), the filter frequency is continuously adjusted in a closed feedback loop with mHz accuracy [20]. The stable and low-noise cw output of the FDML laser, combined with its ability to rapidly sweep through wavelengths, enhances the overall stability of the FWM process. The FDML output was set to a span of 105 nm (Fig. 2, measured with AQ6374, Yokogawa, Japan) with a maximum output power of 15 mW.

 figure: Fig. 2.

Fig. 2. FDML setup on the left with output spectrum on the right. The FDML consists of an amplifier (BOA), a Fabry–Perot bandpass filter (FFP), and a fiber spool as delay. The cFBG is used to counteract dispersion. The sweep range of the output is 105 nm.

Download Full Size | PDF

In this setup, degenerate FWM is used in the PCF (LMA-PM-5, NKT Photonics/Thorlabs) to split the power of the 1064-nm MOPA pump laser into a ∼1300-nm FDML laser seeded idler and a ∼900-nm signal (Fig. 1, red and blue parts). Individual elements of the setup were inspired by Gottschall et al. [9], Nodop et al. [21], and Kolb et al. [22]. Without the FDML seed light, spontaneous FWM results in a broad spectrum at 900 nm and 1300 nm when pumped at 1064 nm. With the 1300-nm FDML seed laser, the efficiency of this process increases, and the signal and idler wavelengths are controlled. The laser beam combining is done in free space using a dichroic mirror (1200 nm Techspec shortpass filter, Edmund Optics, USA). To protect the 1064-nm MOPA laser from back reflections, a 1064-nm free space isolator (IO-5-1064-HP, Thorlabs), a 1064-nm bandpass (L-1064-10, Thorlabs), and a 1000-nm longpass filter (FEL1000, Thorlabs) are used. Different power levels of the 1064-nm MOPA and PCF lengths were tested, resulting in 15 cm of PCF and 30 kW of peak power for optimal results. The ends of the PCF were sealed with a coreless fiber endcap, glued and polished in an FC/APC connector. The high peak power is necessary for the FWM process and to compensate for the losses in the free space setup. At low power levels (no FWM) we measured a throughput of only about 25% of the 1064-nm light after the PCF. This is due to losses in the beforementioned components, which we introduced to prevent back reflections and coupling losses into the PCF. Since the FWM is polarization dependent, the PCF and the 1064-nm MOPA laser are all-PM and an additional half-wave plate (WPH05M-1064, Thorlabs) is used in the free space part to control the polarization. To adjust the polarization of the FDML laser, a polarization controller is placed behind the FDML cavity. After the PCF, all three wavelengths are present (1064 nm, 1300 nm, and 900 nm). For analysis we used spectral filters to select signal, pump, or idler only.

The spontaneous FWM in the LMA-PM-5 when pumped with 1064 nm results in spectrally broad humps around 1300 nm and 900 nm [Fig. 3(A)]. The spectral widths of these humps indicate the gain bandwidth of the FWM setup. The gain profile on the idler side around 1300 nm is well suited for seeding the FWM process with the FDML laser running at 1300 nm with a span of 105 nm (Fig. 2). To showcase the complete tuning bandwidth of the 900-nm spectrum, we intentionally detuned the repetition rates of the MOPA and the FDML laser, which caused the MOPA to amplify each wavelength of the FDML sequentially. As a result, the full 105-nm idler bandwidth of the FDML was amplified, as seen in Fig. 3(B), resulting in a 54-nm span of the signal around 900 nm, as seen in Fig. 3(C). The peaks at the edges of the spectrum are due to the sinusoidal sweep operation of the FDML laser.

 figure: Fig. 3.

Fig. 3. FWM light analysis. (A) Spontaneous FWM when pumped at 1064 nm. (B) Full amplified FDML 1300-nm spectrum. (C) Full 54-nm span of the 900-nm light. (D) 900-nm spectrum with both lasers running at the same frequency. (E) Oscilloscope trace of one 900-nm pulse. (F) Stability of the 900-nm peak power.

Download Full Size | PDF

When both lasers are locked and run with the same repetition rate, only one spectral point of the FDML sweep will be amplified and a corresponding 900-nm spectral point is generated. This results in a spectral FWHM of 70 pm [Fig. 3(D)] in the 900-nm region, which is comparable to fiber optical parametric amplifier setups with Ti:Sa lasers [7] and would result in a spectral resolution of 0.8 cm-1 in stimulated Raman microscopy. When further increasing the 1064-nm pump power, only the spectral FWHM of the 900-nm light increases and not its peak power. This is probably due to SPM and further FWM in the PCF and therefore shifting to other wavelengths. The beginning of this trend can already be seen in Fig. 3(D) at –10 dBm. Shortening the PCF counteracts this and an optimum for this setup was found with 15 cm PCF and 7.5–9 kW pump peak power (30–35 kW 1064-nm peak power out of the laser with 6 dB losses in the free space combiner). The temporal FWHM of the 1064-nm pulse is 100 ps, which is about 20% longer compared to the resulting 79 ps of the 900-nm pulses [Fig. 3(E), measured with DSOZ634A, Keysight, USA and DXM30AF, Thorlabs]. This is due to the nonlinearity of the FWM process and using a quasi-rectangular pulse as input. This results in a pulse energy of about 0.2 µJ of the 900-nm light.

The peak power and pulse energy of this 900-nm light is very stable [Fig. 3(F)]. For characterization, four measurements were taken, each consisting of 10,000 pulses with different pauses in between, to investigate the temporal noise behavior. Initially, every second pulse was measured, then this was increased to every 10th, 100th, and 1,000th pulse. This shows a stable power not only from pulse to pulse, but also over a longer period. The standard deviation of the pulse peak power as well as the pulse energy was always lower than 4%. This is especially desired for nonlinear imaging like TICO-Raman, where a stable tunable light source is necessary [5,9,23].

Since the FDML rapidly tunes its wavelength over time, the wavelength peak of the 1300-nm and 900-nm spectrum can electronically be selected by changing the phase between MOPA and FDML laser, making this a robust and fast way to change the wavelength without the need of a mechanical element. The electronically modulated MOPA laser allows the phase of the pulses to be easily changed, theoretically even from pulse to pulse. To demonstrate the wavelength selection, in Fig. 4 the phase between the lasers was changed five times to amplify and generate five different wavelengths. We see that at all spectral positions a narrowband distinct spectral line is achieved. The difference in maximum amplitude between 900 nm and 1300 nm decreases due to varying chromatic coupling efficiency into the fiber and the spectral profile of the attenuation filter (NENIR40A, Thorlabs). The vendor's data sheet indicates that the filter has a transmission of 0.3% and 0.003% at 900 nm and 1300 nm, respectively. Careful measurements conducted at each wavelength show a peak power split of 21%, 67%, and 12% between 900 nm, 1060 nm, and 1300 nm, respectively.

 figure: Fig. 4.

Fig. 4. Different phase shifts between the FDML and MOPA laser resulting in different amplified and generated light due to the FDML.

Download Full Size | PDF

Considering potential applications of the presented source, we can expect to achieve a wide coverage of the Raman spectrum, when using not only the 900-nm light but also the 1300-nm amplified FDML. Here, 1.4 kW peak power was achieved over 105 nm. Since the power of all three wavelength regions (900 nm, 1064 nm, and 1300 nm) of the laser are sufficient for pumping and probing Raman scattering, combinations of the 900 nm, 1064 nm, and 1300 nm are possible, while being able to sweep rapidly from 875 to 925 nm and 1245 to 1350 nm. This results in a wavenumber coverage from 1360 cm-1 to 2000 cm-1 when using the 1064 nm as pump and the 1245–1350 nm as probe laser. These wavenumbers correspond to parts of the fingerprint region. Also, wavenumbers from 2800 cm-1 to 4000 cm-1 are possible when using the 875–925 nm as pump and the 1245–1350 nm as probe laser, which correspond to the CH-stretch wavenumber region. Therefore, wide parts of the Raman spectrum are covered. The high peak power of all three laser parts could allow for TICO-Raman microscopy with improved noise performance and measurements in epi-direction for future endoscopy applications. We have previously demonstrated that it is possible to efficiently shift the wavelength of our MOPA laser from 1064 nm to 1122 nm and afterwards to 1186 nm with seeded Raman scattering [15]. This could be used to cover Raman bands from 400 cm-1 to 4000 cm-1 [5] by the laser presented in this paper. For the application in SLIDE microscopy, the presented laser will extend the number of accessible fluorophores significantly; especially many additional fluorescent proteins will be usable.

In this paper we demonstrated a stable, rapidly tunable, pulsed-fiber laser based on FWM with an FDML laser at 900 nm and 1300 nm with up to 2.5 kW peak power. The lack of good fiber gain media for high pulse energies in these wavelength regions make this light source especially attractive. The fast tunability of the FDML seed laser and the electronically programmable MOPA pump laser allow for rapid arbitrary selection of wavelengths. This laser will be particularly useful for SLIDE and TICO-Raman microscopy.

Funding

Deutsche Forschungsgemeinschaft (EXC 2167-390884018); Bundesministerium für Bildung und Forschung (BMBF no. 13GW0227B: “Neuro-OCT”); State of Schleswig-Holstein, Germany, (Excellence Chair Program by the Universities of Kiel and Luebeck).

Disclosures

The authors declare no conflicts of interest.

Data availability

Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.

REFERENCES

1. W. Denk, J. Strickler, and W. Webb, Science 248, 73 (1990). [CrossRef]  

2. S. Karpf, C. T. Riche, D. Di Carlo, A. Goel, W. A. Zeiger, A. Suresh, C. Portera-Cailliau, and B. Jalali, Nat. Commun. 11, 2062 (2020). [CrossRef]  

3. D. G. Ouzounov, T. Wang, C. Wu, and C. Xu, Biomed. Opt. Express 10, 3343 (2019). [CrossRef]  

4. Y. Yang, N. Liu, Y. He, Y. Liu, L. Ge, L. Zou, S. Song, W. Xiong, and X. Liu, Nat. Commun. 9, 1504 (2018). [CrossRef]  

5. S. Karpf, M. Eibl, W. Wieser, T. Klein, and R. Huber, Nat. Commun. 6, 6784 (2015). [CrossRef]  

6. N. Dudovich, D. Oron, and Y. Silberberg, Nature 418, 512 (2002). [CrossRef]  

7. M. Chemnitz, M. Baumgartl, T. Meyer, C. Jauregui, B. Dietzek, J. Popp, J. Limpert, and A. Tünnermann, Opt. Express 20, 26583 (2012). [CrossRef]  

8. C. W. Freudiger, W. Yang, G. R. Holtom, N. Peyghambarian, X. S. Xie, and K. Q. Kieu, Nat. Photonics 8, 153 (2014). [CrossRef]  

9. T. Gottschall, T. Meyer-Zedler, M. Schmitt, R. Huber, J. Popp, A. Tünnermann, and J. Limpert, J. Raman Spectrosc. 52, 1561 (2021). [CrossRef]  

10. D. Wu, Z. Guo, Z. Wu, and P. P. Shum, J. Opt. 20, 035501 (2018). [CrossRef]  

11. L. Lavoute, J. C. Knight, P. Dupriez, and W. J. Wadsworth, Opt. Express 18, 16193 (2010). [CrossRef]  

12. R. Huber, M. Wojtkowski, and J. G. Fujimoto, Opt. Express 14, 3225 (2006). [CrossRef]  

13. T. Pfeiffer, M. Petermann, W. Draxinger, C. Jirauschek, and R. Huber, Biomed. Opt. Express 9, 4130 (2018). [CrossRef]  

14. D. C. Adler, Y. Chen, R. Huber, J. Schmitt, J. Connolly, and J. G. Fujimoto, Nat. Photonics 1, 709 (2007). [CrossRef]  

15. S. Karpf, M. Eibl, B. Sauer, F. Reinholz, G. Hüttmann, and R. Huber, Biomed. Opt. Express 7, 2432 (2016). [CrossRef]  

16. M. Eibl, S. Karpf, H. Hakert, T. Blömker, J. P. Kolb, C. Jirauschek, and R. Huber, Opt. Lett. 42, 4406 (2017). [CrossRef]  

17. P. J. Mosley, S. A. Bateman, L. Lavoute, and W. J. Wadsworth, Opt. Express 19, 25337 (2011). [CrossRef]  

18. T. Gottschall, T. Meyer, M. Baumgartl, C. Jauregui, M. Schmitt, J. Popp, J. Limpert, and A. Tünnermann, Laser Photonics Rev. 9, 435 (2015). [CrossRef]  

19. C. Grill, T. Blömker, M. Schmidt, D. Kastner, T. Pfeiffer, J. P. Kolb, W. Draxinger, S. Karpf, C. Jirauschek, and R. Huber, Commun. Phys. 5, 212 (2022). [CrossRef]  

20. S. Lotz, C. Grill, M. Göb, W. Draxinger, J. P. Kolb, and R. Huber, Biomed. Opt. Express 12, 2604 (2021). [CrossRef]  

21. D. Nodop, C. Jauregui, D. Schimpf, J. Limpert, and A. Tünnermann, Opt. Lett. 34, 3499 (2009). [CrossRef]  

22. J. P. Kolb, W. Draxinger, D. Weng, H. Hakert, S. N. Karpf, R. Huber, J. Popp, T. Meyer, J. Limpert, T. Gottschall, M. Eibl, R. Brinkmann, and R. Birngruber, in Multiphoton Microscopy in the Biomedical Sciences XIX, A. Periasamy, P. T. So, and K. König, eds. (SPIE, 2019), p. 83.

23. H. Hakert, M. Eibl, M. Tillich, R. Pries, G. Hüttmann, R. Brinkmann, B. Wollenberg, K.-L. Bruchhage, S. Karpf, and R. Huber, Opt. Lett. 46, 3456 (2021). [CrossRef]  

Data availability

Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.

Cited By

Optica participates in Crossref's Cited-By Linking service. Citing articles from Optica Publishing Group journals and other participating publishers are listed here.

Alert me when this article is cited.


Figures (4)

Fig. 1.
Fig. 1. FWM setup. Green: 1064-nm MOPA laser used as pump. EOM, electro-optic modulator; YDFA, ytterbium-doped fiber amplifier. Red: free space combining of the 1064-nm MOPA laser and the 1300-nm FDML laser. BP, bandpass filter; LP, longpass filter; SP, shortpass filter. Blue: FWM in the PCF with filters afterwards.
Fig. 2.
Fig. 2. FDML setup on the left with output spectrum on the right. The FDML consists of an amplifier (BOA), a Fabry–Perot bandpass filter (FFP), and a fiber spool as delay. The cFBG is used to counteract dispersion. The sweep range of the output is 105 nm.
Fig. 3.
Fig. 3. FWM light analysis. (A) Spontaneous FWM when pumped at 1064 nm. (B) Full amplified FDML 1300-nm spectrum. (C) Full 54-nm span of the 900-nm light. (D) 900-nm spectrum with both lasers running at the same frequency. (E) Oscilloscope trace of one 900-nm pulse. (F) Stability of the 900-nm peak power.
Fig. 4.
Fig. 4. Different phase shifts between the FDML and MOPA laser resulting in different amplified and generated light due to the FDML.
Select as filters


Select Topics Cancel
© Copyright 2024 | Optica Publishing Group. All rights reserved, including rights for text and data mining and training of artificial technologies or similar technologies.