Toward a 10-W level femtosecond erbium laser operating in the 1600 nm low gain region

The 1.6 μm laser wavelength is in the near-infrared-II (NIR-II) biological window, which is very useful for deep multiphoton microscopy due to its lower scattering loss compared with a typical NIR-I biological window from 0.76 to 0.9 μm.^1,2 The frequency-doubled light of an ultrafast 1.6 μm all-fiber laser is at ∼800 nm, which can also serve as a substitute for bulky Ti–sapphire lasers. For both erbium-doped fibers (EDFs) and thulium-doped fibers (TDFs), the 1.6 μm waveband is located in a low gain region,^3,4 so realizing a high power 1.6 μm laser is difficult. For continuous wave (CW) 1.6 μm laser amplification, increasing the EDF length is an effective method to provide a sufficient gain for the 1.6 μm waveband^5–8 because EDFs have a high re-absorption at the C-band so that lengthening the gain fiber could consume the C-band component, which, in turn, pumps the remaining Er³⁺ to increase the 1.6 μm gain. Various pumping schemes, such as auxiliary C-band pumping,^9–12 in-band pumping,^13,14 and 1018 nm laser pumping,^15–17 were applied in long EDF amplifiers to further scale up the power of CW 1.6 μm lasers. These approaches are not suitable for ultrafast laser amplification because a long fiber will accumulate a severe nonlinearity. Designing an amplifier for a 1.6 μm ultrafast laser encounters difficulties in managing nonlinearity in the amplifier. Obtaining a stable 1.6 μm ultrafast seed is also not easy. A solid-state optical parameter oscillator (OPO) is an alternative to obtain a high-power laser at the 1.6 μm waveband¹⁸ but is so bulky.

Many high power CW 1.6 μm lasers have already been reported.^10–17 However, there are few reports on high power 1.6 μm ultrafast lasers. Franck Morin et al. achieved an ∼605 fs, ∼460 mW, 1.6 μm SSFS laser in a large-mode-area (LMA) amplifier.¹⁹ However, the laser power of the SSFS cannot increase continuously with the pump power. In 2018, Kang et al. built a chirped pulse amplification (CPA) system for a 1.6 μm ultrafast laser, which could produce an output power of ∼167 mW, corresponding to a pulse energy of ∼14 nJ. However, 22% of the total amplified power was occupied by the C-band amplified spontaneous emission (ASE).²⁰ In our previous study,²¹ we increased the 1.6 μm dissipative soliton (DS) laser power from 1 to 2 W by managing the gain fiber length, but the laser pulses degraded into a noise-like state in the long gain fiber.

In this paper, we construct a high power 1.6 μm fs CPA system. The CPA system contains five parts, which are a chirped 1.6 μm soliton-self-frequency-shift (SSFS) seed, three-stage pre-amplifiers, a double-pass amplifier, a C-band assisted LMA main amplifier, and a pulse compressor. The double-pass amplifier provides a watt-level injection laser power for the final main amplifier. To alleviate the nonlinearity in the main amplifier, the seed is highly chirped and the gain fiber in the main amplifier has a large-mode area. The auxiliary laser wavelength needs to be adjusted to correspond to the length of the LMA gain fiber. A tunable C-band (1530–1560 nm) laser allows flexible wavelength adjustment of the auxiliary laser. For an ∼5 m LMA gain fiber, a maximum output power of ∼9.42 W is achieved when injecting an ∼500 mW, 1550 nm laser as the auxiliary laser. For an ∼3 m LMA gain fiber, the auxiliary laser wavelength needs to be re-optimized because the absorption of the gain fiber has changed after reducing the length. The maximum output power achieved in the main amplifier with this shorter LMA gain fiber is ∼9.30 W. Considering the relatively long laser chain we built, the 1.6 μm seed should be highly chirped to a picosecond (ps). The high power ps-scale 1.6 μm laser after passing the LMA main amplifier is compressed to the fs-scale using a free-space grating pair. Compared with previous research about CW 1.6 μm lasers, this CPA laser system with the C-band auxiliary laser pumping scheme allows a much shorter gain fiber length to achieve a high power 1.6 μm amplification, which is more suitable for amplifying ultrafast 1.6 μm lasers.

Figure 1 shows the experimental setup of the high power 1.6 μm fs CPA system. The ultrafast 1.6 μm seed is generated by the SSFS technique based on a 1.56 μm femtosecond laser. The ∼36 MHz, 1.56 μm femtosecond laser is amplified using a piece of EDF (LIEKKI Er110-4/125) pumped using a 976 nm laser diode (LD) and then propagates through an ∼10 m SM1950 (Nufern) fiber to trigger a laser wavelength shift. The spectrum after the SM1950 fiber is shown in Fig. 1(b). A fiber filter (filter 1) is placed behind the SM1950 fiber to block the C-band component to get a pure ultrafast 1.6 μm seed. The 1.6 μm fs seed is stretched using an ∼55 m dispersion compensation fiber (DCF, DM1011-A, YOFC). The highly chirped 1.6 μm seed is further amplified using three pre-amplifiers. Between each amplifier, a C-band fiber filter (filter 2, filter 3, filter 4, or filter 6) guarantees a pure 1.6 μm amplified laser before the next amplifier. The injection power of the 1.6 μm laser is very important for each amplifier. If the injection power is too weak, the 1.6 μm signal laser cannot compete with the C-band ASE, which makes the 1.6 μm laser power saturate quickly. This is why there are many cascade amplifiers used in our system. After the three pre-amplifiers, the amplified pure 1.6 μm laser is injected into the double-pass amplifier constructed using an ∼3.5 m Er:Yb co-doped double-cladding fiber (Coractive DCF-EY-10/128H), through a circulator from port 1 to port 2. The double-pass amplifier can amplify the 1.6 μm laser directly to the watt-level power with a high suppression of the C-band ASE. In the double-pass amplifier, port 1 and port 3 of the circulator are spliced together to form a loop. In the loop, a fiber filter (filter 5) is used to remove the C-band ASE of the amplified forwarding laser. Using a wave division multiplexer (WDM), the C-band auxiliary CW laser is injected into the LMA main amplifier with the watt-level 1.6 μm laser.

The passive fiber used in the main amplifier is an LMA fiber (IXF-2CF-PAS-25-250-0.08) having an ∼25 μm core diameter. The spectra of the C-band auxiliary lasers are shown in Figs. 1(c)–1(f). The gain fiber (IXF-2CF-EY-O-25-250-COM) has the same core diameter as the passive fiber. The LMA gain fiber has the highest absorption of 76 dB/m at ∼1530 nm. The high absorption loss at the C-band suppresses the C-band ASE and guarantees sufficient absorption of the auxiliary laser. The output laser is collimated using a plano–convex mirror (M1). The ps-scale high power 1.6 μm laser is compressed using a free-space grating pair. M2 is a long-pass filter (LPF, No. 84672, Edmund Optics) that blocks most of the residual C-band laser.

The power of the laser injected into the LMA main amplifier should be maximized to suppress the C-band ASE. For this goal, we investigated two structures. Figure 2(a) shows the output power of the single-pass structure. After the C-band fiber filter (filter 6), the pure 1.6 μm laser saturates when the pump power increases to ∼10 W. The laser power is clamped at ∼1.2 W. From the measured spectra depicted in Fig. 2(b), the C-band ASE appears at the pump power of ∼10 W. At the maximum pump power of ∼17 W, the intensity of the C-band ASE is comparable to the 1.6 μm laser, which indicates that most of the pump power contributes to the C-band ASE. In the double-pass structure, the C-band ASE in the amplified forwarding laser is blocked using the fiber filter (filter 5) embedded in the loop constructed using the circulator. The returned pure amplified 1.6 μm laser is amplified using the Er:Yb co-doped double-cladding fiber again. At the maximum pump power of ∼17 W, the C-band ASE is suppressed better than in the single pass structure. The intensity of the C-band ASE is ∼25.138 dB lower than that of the amplified 1.6 μm laser, as shown in Fig. 2(d). After the fiber filter (filter 6), the highest pure amplified 1.6 μm laser power is ∼2.2 W, with a slope efficiency of ∼13.8%. The power increase trend in the double-pass structure is linear, and no power saturation is observed. The pulse duration of the double-pass structure at the maximum pump power of ∼17 W is ∼106.540 ps.

In general, the main emission of the Er³⁺-doped fiber is concentrated in the C-band wavelength range of 1530–1565 nm. The maximum emission cross section is at ∼1530 nm. Due to the high re-absorption of the C-band wavelength, the ∼1560 nm wavelength has a maximum net gain. This explains why an erbium laser works at ∼1560 nm although the maximum emission cross section is at ∼1530 nm. As shown in Fig. 3, the particles in the ground state absorb the pump energy to jump to the upper state and then decay to the ⁴I_13/2 energy level. The population inversion between ⁴I_13/2 and the ground state corresponds to a laser emission at a C-band wavelength. The C-band and 1.6 μm emissions share one ⁴I_13/2 energy level splitting into two sub energy levels, which are ⁴I_13/2b and ⁴I_13/2a. The 1.6 μm emission occurs between ⁴I_13/2b and the ground state. Taking advantage of the high re-absorption of the EDF at the C-band wavelength, a long active fiber will absorb the C-band laser, achieving rapid in-band relaxation between ⁴I_13/2b and ⁴I_13/2a. The ⁴I_13/2b energy level acts as a reservoir, which provides particles for ⁴I_13/2a to achieve population inversion between ⁴I_13/2b and the ground state.

In the LMA main amplifier, we first used an ∼5 m LMA gain fiber. Three wavelengths were chosen as the C-band auxiliary lasers: 1530, 1550, and 1560 nm. Since the amplified 1.6 μm laser has a lower gain coefficient than the auxiliary laser, the 1.6 μm injection laser should have a high power to provide a dominant position when competing with the auxiliary laser for gain. The laser power injected into the LMA main amplifier was ∼2.2 W. The spectra of the 1530, 1540, 1550, and 1560 nm auxiliary lasers are shown in the insets of Figs. 1(c)–1(f), respectively. Figure 4(a) shows the output power from the LMA amplifier without the auxiliary C-band laser. The maximum output power is ∼7.52 W when the 915 nm pump power is ∼35 W. The slope efficiency is ∼18.12%, but the power increase is slightly non-linear.

As shown in Fig. 4(b), no C-band ASE is observed, and the residual 915 nm pump can be ignored.

Figure 5(a) shows the output power of the main amplifier with the seeding of the 1530, 1550, and 1560 nm auxiliary lasers with the same power of ∼167 mW. After seeding these auxiliary lasers, the output power increases from the original 7.52 to ∼8.56, ∼8.75, and ∼8.47 W, respectively. The corresponding slope efficiencies are ∼20.7%, ∼21.5%, and ∼19.97%. Figures 5(b)–5(d) show the amplified spectra for seeding with different auxiliary laser wavelengths. The EDF has the highest absorption at ∼1530 nm; therefore, the 1530 nm auxiliary laser would be rapidly consumed when it propagates through the gain fiber with the high power 1.6 μm laser. Thus, the 1530 nm auxiliary laser is almost entirely absorbed by the gain fiber, as shown in Fig. 5(b). The C-band ASE in Fig. 5(b) is mainly coming from the initial ∼169.7 mW 1530 nm auxiliary laser [Fig. 1(c)] and is then amplified using the LMA gain fiber. Since the C-band ASE is mainly concentrated in the 1560–1570 nm region, which cannot be absorbed fully by the gain fiber, the energy associated with this portion does not participate in amplifying the 1.6 μm laser. Fortunately, the amplified C-band ASE is still much lower than the 1.6 μm component, so it can be ignored. For seeding the 1550 nm auxiliary laser, the intensity ratio is over 25 dB between the 1550 nm auxiliary laser and the 1.6 μm laser, as shown in Fig. 5(d). The maximum output power of ∼9.42 W is achieved by using 1550 nm as the auxiliary laser wavelength. Since the EDF has a lower absorption (but the maximum net gain) at 1560 nm, the main process that the 1560 nm auxiliary laser undergoes is amplification rather than absorption. Thus, the final amplified spectrum of the 1.6 μm laser contains a significant residual amplified 1560 nm component, as shown in Fig. 5(c).

As shown in Fig. 5, the 1530 nm and 1550 nm auxiliary laser wavelengths have better outcomes. In Figs. 6(a) and 6(b), we further investigate the 1530 and 1550 nm auxiliary laser powers seeded into the LMA amplifier. When seeding the ∼399 mW 1530 nm laser, the output power can reach 8.67 W with a slope efficiency of ∼21.25%, while seeding the ∼500 mw 1550 nm laser, the maximum output power can reach 9.42 W and the slope efficiency is up to ∼23.23%. As shown by the spectra in Fig. 6(c), the 1.6 μm laser component is still much higher (>20 dB) than the residual 1550 nm auxiliary laser. The highly chirped 9.42 W, 1.6 μm laser is compressed using a free-space grating pair. The achieved narrowest pulse duration is ∼361 fs. The whole compression efficiency is ∼77%; the laser power after the compressor is ∼7.25 W, corresponding to a 201 nJ pulse energy and a 557 kW peak power. The long-term power stability over 2 h is shown in Fig. 6(e). The root mean square (RMS) of the power fluctuation is ∼0.98%. The ASE in the 1530 nm auxiliary laser is the main factor resulting in a lower amplifier power compared with the 1550 nm auxiliary laser. The intrinsic mechanism of the C-band auxiliary laser pumping is that the auxiliary laser is amplified first to provide enough pump power for the 1.6 μm laser and is then absorbed by the long gain fiber to increase the 1.6 μm laser power. The 1530 nm is not the strongest emission of the EDF, so it cannot get sufficient power amplification. In the amplification process, most pump power contributes to the 1560–1570 nm ASE carried using the initial 1530 nm auxiliary laser. Due to the large ps-scale pulse duration and the use of the LMA gain fiber, the nonlinearity in the amplification is controlled well and no obvious broadening occurs on the amplified spectrum.

For typical ultrafast laser amplification, a short gain fiber is required to avoid the severe nonlinearity. Thus, we decreased the LMA gain fiber length to ∼3 m. A shorter gain fiber means a lower absorption at 1550 and 1560 nm. Thus, for the 3 m LMA amplifier, we used 1530 and 1540 nm as the auxiliary laser wavelengths. After seeding the ∼274.5 mW 1540 nm auxiliary laser [Fig. 1(d)] and the ∼402 mW 1530 nm auxiliary laser, the output power increased to ∼9.3 and ∼9.2 W, respectively. The corresponding slope efficiencies are ∼22.606% and ∼21.939%, as shown in Fig. 7(a). Since the absorption decreases with decreasing gain fiber length, the residual auxiliary laser and the residual pump power are higher than those from the ∼5 m LMA gain fiber. For this situation, seeding the 1530 nm auxiliary laser is better than seeding the 1540 nm auxiliary laser. The 1.6 μm laser component is more than 20 dB higher than the residual 1530 nm auxiliary laser and the ASE, while it is only ∼17.554 dB higher when using the 1540 nm as the auxiliary laser wavelength. Both the output power and the slope efficiency achieved in the shorter main amplifier are lower than in the ∼5 m main amplifier, but a shorter main amplifier is more suitable for further enhancing the ultrafast power beyond 10 W because a shorter gain fiber means a lower nonlinearity.

A comparison of the previously reported high power 1.6 μm lasers is shown in Table I. One can see that this work demonstrates the highest power 1.6 μm laser by using an all-fiber CPA system and an auxiliary C-band laser co-pumping scheme. In the actual experiment, the power saturation is not observed, indicating that the output power can be further increased by using a higher pump power. If a higher pump power is used, the laser power can be expected to be greater than 30 W. Adopting the gain fiber with a larger mode area, the C-band laser co-pumping scheme has potential to achieve tens or even hundreds of watts. With the development of an S-band laser, whether as an auxiliary laser or as a sole pump source, a high power 1530 nm laser will be the optimal choice for generating a higher power 1.6 μm ultrafast laser.

In conclusion, we designed a hybrid amplifier consisting of a double-pass amplifier and a C-band laser assisted LMA main amplifier for amplifying a weak 1.6 μm ultrafast laser to several watts. In the ∼5 m LMA amplifier, the best auxiliary laser wavelength is 1550 nm, the maximum output power can reach ∼9.42 W, and the slope efficiency is ∼23.26%. The high power 1.6 μm ultrafast laser has a large ps-scale duration, which is further compressed to ∼361 fs after passing a free-space grating pair. For the 3 m short LMA amplifier, by optimizing the auxiliary laser wavelength (using the 1530 nm and 1540 nm lasers as the auxiliary lasers), the maximum output power is ∼9.30 W, corresponding to a slope efficiency of ∼22.606%. Our work provides a new perspective for obtaining an ultra-high power fs 1.6 μm all-fiber laser for, , for example, deep imaging of biological tissues.

We acknowledge the financial support provided by the “Pioneer” and “Leading Goose” R&D Program of Zhejiang Province (Grant No. 2023C03083), in part by the National Key Research and Development Program of China (Grant No. 2022YFC3601003), in part by the National Natural Science Foundation of China (Grant Nos. W2412107, 91833303, and 11621101), in part by the Ningbo Science and Technology Project (Project Nos. 2023Z179, 2021Z030, and 2018B10093), and in part by the Ningbo Public Welfare Research Program (Project No. 2024Z234). The authors are also grateful to Dr. Julian Evans of Zhejiang University for a helpful discussion.

The authors have no conflicts to disclose.

Haolin Yang: Data curation (equal); Investigation (equal); Methodology (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). Ruili Zhang: Funding acquisition (supporting). Sailing He: Funding acquisition (equal); Investigation (equal); Methodology (equal); Supervision (lead); Writing – review & editing (equal).

The data that support the findings of this study are available from the corresponding author upon reasonable request.