1. Introduction

Thulium-doped fiber lasers (TDFLs) are emerging as excellent sources of mid-infrared radiation for various applications, including spectroscopy, sensing, and medicine [1–3]. The 2 μm wavelength is particularly desirable in many medical procedures, thanks to very strong tissue absorption in this spectral region. In consequence, the penetration depth of 2 μm radiation is much smaller in comparison to shorter wavelengths. It has been already proven, that 2 μm sources outperform 1.55 and 1.06 μm lasers in e.g. ablation of urinary tissues [4,5]. The second application, where Tm-doped fiber lasers might be used is laser spectroscopy and trace-gas sensing. The spectral region of 1.9 – 2.0 μm (which is ideally covered by the emission of Tm-lasers) contains several strong absorption lines of various molecules, especially two greenhouse gases: carbon dioxide (CO₂) and nitrous oxide (N₂O) [6]. Thus, research on compact, fully-fiberized broadband TDFLs (i.e. mode-locked all-fiber lasers) might essentially contribute to the development of ultra-sensitive, robust and portable trace-gas detection systems. Moreover, the main advantage of mode-locked lasers, namely the pulse duration at the level of femtoseconds, makes those lasers much more useful in many applications.

There were several reports on fully fiberized, ultrashort pulse Tm-doped lasers in the literature. So far, passively mode-locked lasers based on semiconductor saturable absorber mirrors (SESAMs) [7], graphene [8–12], carbon nanotubes [13], and topological insulator [13] were demonstrated. However, in all those cases the average power and pulse energy were strongly limited, either by the SA damage threshold or by the nonlinear effects in the fiber-based resonator. Among the mentioned works in all-fiber lasers, the highest energy was achieved by M. Chernysheva et al. [13] (0.4 nJ and 18 mW of average power). Larger pulse energy might be obtained with proper intra-cavity dispersion management, since the soliton area theorem limits the maximum pulse energy of an all-anomalous dispersion laser. Other dispersion regimes (like all-normal or near-zero) enable achieving larger pulse energies. This can be done by applying e.g. a grating-based compressor inside the cavity [14–16]. However, such approach is very inconvenient, since Martinez-type compressors require careful adjustment and introduce large losses. That is why the setups require very strong multi-watt pumping sources [14,15]. The dispersion might be compensated using proper optical fibers. One can force the laser to operate in the normal net dispersion regime and generate dissipative solitons. This approach led to generation of 130 fs pulses with impressive 7.6 nJ energy [17]. However, the setup required the usage of 2 types of dispersion compensating fibers (ultra high numerical aperture fibers, UHNA) and complex free-space optics set responsible for the nonlinear polarization evolution (NPE) mode-locking (a Lyot filter together with waveplates, polarizing cubes, etc.). Additionally, the pulses were dechirped outside the cavity using a bulk grating compressor. Alternatively, the dispersion of the oscillator might be balanced. A Holmium-doped fiber laser with zero net group delay dispersion (GDD) was demonstrated by P. Li et al. [18]. It delivered 160 fs pulses with 1 nJ energy at 2060 nm. However, the setup was also based on free-space optics, which requires adjustment (three waveplates, a polarizing beamsplitter and isolator). Very similar setup, but with the use of Tm-doped fibers was presented by A. Wienke et al. [19]. The laser operated in the stretched-pulse regime (near-zero GDD) and delivered very short 119 fs pulses, but the energy was at the level of 0.16 nJ.

Here, we demonstrate a simple, compact and fully fiberized laser system, which delivers 260 fs-short pulses at 1968 nm wavelength with energy exceeding 1 nJ. We have shown experimentally, that nJ-level pulses in the 2 μm region might be obtained not only with complex free-space optical systems or grating-based compressors, but also with the use of simple, all-fiber solutions.

2. Experimental setup

The experimental setup of the laser system is depicted in Fig. 1. It comprises an oscillator and all-fiber CPA. The seed laser consists of: a hybrid component comprising an 10% output coupler (OC), wavelength-division multiplexer (WDM) and isolator (ISO), a piece of Tm-doped fiber (TDF), a polarization controller (PC) and the graphene-based saturable absorber (GSA). The hybrid device ensures unidirectional light propagation in the cavity (marked counter-clockwise on the schematic in Fig. 1) and reflects a 10% fraction of intra-cavity power to the output port. Thanks to the use of the hybrid component, the number of components in the laser cavity was minimized, which ultimately reduces the overall complexity of the system. As active medium, a short piece (approx. 16 cm) of Nufern TSF-5/125 Tm-doped fiber was used. The oscillator is pumped by a fiber laser source operating at 1566 nm and generating up to 400 mW of power. The total cavity length is approx. 2.03 m, which corresponds to 100.25 MHz repetition frequency.

 

Fig. 1 Experimental setup of the Tm-doped fiber laser system.

Download Full Size | PDF

The pulses from the oscillator are stretched in a segment of dispersion compensating fiber (DCF) with mode field diameter of 5.8 μm (at 2000 nm) and group velocity dispersion (GVD) of approx. 0.033 ps²/m at 2000 nm. In our experiment, the stretcher introduces GDD of 3 ps². Temporarily broadened pulses are afterwards amplified in a fiber amplifier, which comprises the same TDF type as the oscillator (30 cm length). The active fiber is bi-directionally pumped at 1566 nm wavelength using a laser diode amplified in a self-made Erbium-Ytterbium co-doped fiber amplifier [20] (with maximum available power of 2.46 W) via filter-type WDMs (FWDM), which reflect the pump light. After amplification, the pulses are compressed in a 20.4 meter long segment of standard single-mode fiber (SMF) with GDD of 0.083 ps²/m at 2000 nm [15]. In order to avoid any back-scattered light, two isolators were placed at both ends of the amplifier. The proposed architecture requires usage of several fiber-based components (which are commercially available), however, there are no free-space coupled, bulk elements like polarizers, wave-plates, lenses, etc. Thus, the whole system is invulnerable to external disturbances and does not require any alignment.

The saturable absorber used in the oscillator is based on a 12-layer graphene/polymer composite. The graphene layers were grown by chemical vapor deposition (CVD) on copper substrate using Aixtron Black Magic Pro system and afterwards immersed in a poly(methylmethacrylate) (PMMA) support, forming a stable, free-standing foil containing 12 graphene layers (a photograph of the fabricated sample is shown in Fig. 2(b)). The details on the graphene/PMMA composite fabrication process were described previously in [12]. A small piece of the composite (approx. 0.8x0.8 mm size) was cut from the sample by a sharp blade and afterwards placed on the end facet of a fiber connector, and connected with a clean one. Such device is then ready to use as SA in a fiber laser. The nonlinear optical properties of the fabricated SA were examined in a power-dependent transmission measurement setup, similar to that presented previously in [21], typically used for characterization of fiber-based saturable absorbers. Due to lack of a proper pumping source and enough sensitive photodetectors for 2 μm wavelength, the SA was tested using 1560 nm laser. As an excitation source, a pulsed laser was used, with 2 ps pulse duration and 100 MHz repetition rate. The maximum achievable fluence was at the level of 500 μJ/cm². The power-dependent transmission curve with indicated parameters (modulation depth ΔT, saturation fluence F_sat, non-saturable losses α_NS) together with theoretical fitting are plotted in Fig. 2(a).

 

Fig. 2 Saturable absorption curve of the 12-layer graphene composite measured at 1560 nm (a), broadband absorption of the SA in the linear regime (b). Inset: photograph of a 12 layer graphene/PMMA foil.

Download Full Size | PDF

The fit was calculated using the following formula, valid for fast saturable absorbers [22–24]:

The broadband absorption in the linear regime, measured with the use of a white light source and optical spectrum analyzer is plotted in Fig. 2(b). It can be seen, that the transmission of the SA is almost equal at both 1.56 and 1.97 μm wavelengths (~71%). The transmittance drops above 2000 nm, most likely due to the attenuation of PMMA polymer in the mid-infrared region.

3. Experimental results

The performance of the laser was characterized using an optical spectrum analyzer (Yokogawa AQ6375), radio frequency (RF) spectrum analyzer with 3.6 GHz bandwidth (Keysight EXA N9010A) coupled with a 16 GHz photodiode (Discovery Semiconductors DSC2-50S), and an autocorrelator (Femtochrome FR-103XL).

The pulses from the seed oscillator are temporarily broadened, amplified and compressed in a simple, all-fiber CPA scheme. Figure 3 depicts the optical spectra observed in different points in the system. The spectrum generated by the seed is shown in Fig. 3(a). It is centered at 1968 nm and has a full width at half maximum (FWHM) of 9.4 nm. The oscillator starts to operate in the mode-locked regime by itself by launching the pump power above a threshold of 280 mW. The pump power might be increased up to 360 mW without losing the mode-locking. Further increase of the pump power leads to parasitic CW lasing, which co-exists with the mode-locking. Hence, all measurements were performed at 360 mW of pumping. At this pump power level, the average output power and the pulse energy from the oscillator were of 8 mW and ~80 pJ, respectively. The repetition rate, resulting from the cavity length is equal to 100.25 MHz (see RF spectrum in Fig. 4). The signal to noise ratio (SNR) in the RF spectrum measured with 33 Hz resolution bandwidth (RBW) is at the level of 75 dB, which is better in comparison to other graphene-based TDFLs reported previously [8–12]. The spectrum measured in the full available frequency span (shown as inset graph in Fig. 4) does not contain any modulations, which confirms stable, single-pulse operation (any multi-pulsing or harmonic mode-locking is easily observable in the RF spectrum, see e.g [25].).

 

Fig. 3 Optical spectra observed at different stages: (from top to bottom) oscillator, after stretching in the DCF fiber, after amplification and after recompression in the SMF.

Download Full Size | PDF

 

Fig. 4 a) Measured RF spectrum of the oscillator, b) autocorrelation traces of the seed (blue dashed-dotted line) and the amplified pulse (red line).

Download Full Size | PDF

The pulses from the oscillator are stretched to about 12 ps in a 90-meter long segment of DCF with group velocity dispersion (GVD) at the level of 0.033 ps²/m at 1968 nm. The total insertion losses of the stretcher (including splice loss and fiber attenuation) are slightly less than 3 dB. The optical spectrum of the pulses after passing through the stretcher is depicted in Fig. 3(b). The spectrum slightly broadens to 15 nm of FWHM. The spectrum plotted in Fig. 3(c) is taken from the 1% coupler port after amplification (prior to compression) at maximum available pump power. It can be seen, that the amplifier does not change the shape of the spectrum and the 15 nm bandwidth is still maintained. The additional small peaks that can be observed in the spectra after stretching and amplification (between the 1st and 2nd sideband) most likely originate from four-wave mixing (FWM) between Kelly’s sidebands. The amplified pulses are then compressed in a 20.4 meter long segment of SMF. The length of the anomalous dispersion SMF was carefully optimized (experimentally) in order to compensate the normal dispersion of the DCF and achieve the shortest pulses at the output under maximum pumping level. The optical spectrum after passing through the SMF is depicted in Fig. 3(d). Small distortions in the spectrum due to nonlinear effects can be observed, however, the ripples in the spectrum do not exceed 2 dB. The pulse duration after compression is 260 fs (assuming a sech² pulse shape typical for solitons with 1.54 deconvolution factor) – the measured autocorrelation is depicted in Fig. 4 (red line) together with the initial pulse from the oscillator (dashed-dotted blue line). The recompressed pulse contains a small pedestal, which might originate from the nonlinearities arising in the compression fiber, however, it contains less than 9% of the total pulse energy. The autocorrelation was measured with two different scan ranges: 4 ps and 16 ps (inset graph in Fig. 4). It can be seen, that the pulse was not only significantly amplified, but its duration was reduced by a factor of 2.5. The 16 ps scan proves, that the output pulse train does not contain any pre- or post-pulses. With 110 mW of average power and 260 fs pulse duration at 100 MHz repetition rate, the pulse energy can be calculated as 1.1 nJ. The peak power, taking into account the energy stored in the side wings, reaches 4 kW.

4. Summary

Summarizing, we have experimentally demonstrated generation of 260 fs-short pulses from a fully fiberized, mode-locked Tm-doped fiber laser with energy and peak power exceeding 1 nJ and 4 kW, respectively. The all-fiber, monolithic system comprises a simple Tm-doped oscillator mode-locked by graphene-based SA, and a Tm-doped fiber amplifier. The whole design is based on single-mode fibers and fiber-based components, without any free-space optics. Such simple CPA system generates 110 mW of average power at the repetition rate of 100 MHz and central wavelength of 1968 nm. This is, to our knowledge, the highest pulse energy generated from a fully fiberized, femtosecond Tm-doped laser. Such compact, robust and cost-effective system might serve as a pump source for nonlinear frequency conversion, mid-IR supercontinuum generation, or as laser source for compact and portable trace-gas detection platforms.