1. Introduction

Wavelength-tunable $Q$-switched fiber lasers have attracted much attention due to their intensive applications in communications, reflectometry, and fiber optical sensing. Compared to an actively $Q$-switched fiber laser, a passive one with a saturable absorber (SA) could be preferred because of the advantages of compactness, low cost, and flexibility in design.

In recent years, passively $Q$-switched fiber lasers have been widely reported using several kinds of SAs, e.g., semiconductor saturable-absorption mirrors [1], carbon nanotubes [2], topological insulators [3,4], and graphene [5–27]. Among these SAs, graphene, an ultrabroadband wavelength-independent saturable-absorption material, has been well recognized for realizing wavelength widely tunable $Q$-switched fiber lasers, especially in erbium-doped fiber (EDF) lasers [5–8]. At present, the largest tunable range of 50.5 nm was reported from a graphene-based $Q$-switched EDF laser using a tunable band-pass filter [5]. However, because graphene has a relatively low saturating intensity and low damage threshold, $Q$-switching operation in a graphene-based fiber laser could disappear under only moderate pumping intensity, also strongly restricting the available pulse energy. By surveying those graphene-based $Q$-switched all-fiber lasers reported previously, one can find the record pulse energy is only 184 nJ [9]. To scale up the $Q$-switched pulse energy and the damage power of graphene SA, some researchers have recently attempted to expand the laser spot for purposely declining the optical intensity injected into graphene SA. The use of such a method has successfully obtained a pulse energy as high as 18 μJ and an average output power of 5.1 W from a graphene-based $Q$-switched ${\mathrm{Tm}}^{3+}$-doped fiber laser [10], but it suffers from the bulky construction and requires complex coupling alignment from free-space into small-core fiber. Therefore, there are always strong motivations to develop large-energy, all-fiber tunable lasers $Q$-switched by graphene SA.

On the one hand, most previous works [8,10–15] have used multilayer graphene by liquid-phase exfoliation or the chemical vapor deposition (CVD) method. Although multilayer graphene could partially enhance the absolute modulation depth of the whole SA, larger nonsaturable loss is induced and easily accumulates more thermal energy in the SA, leading to a lower damage threshold and finally limiting the available pulse energy. In contrast, mono-layer graphene as a SA will provide the following advantages: (1) very low nonsaturable loss ($\sim 2.3\%$ [28]) is favorable to a relatively higher damage threshold, and (2) the use of a mono-layer structure could be better to reflect the intrinsic optical properties or physics of graphene [29]. On the other hand, instead of the conventional single-mode EDF used in some graphene-based $Q$-switched fiber lasers reported previously [5–8,12–22], Er:Yb-codoped double-clad fiber (EYDCF) as the laser gain medium could be more preferable because (1) Er:Yb-codoped fiber possesses higher absorption/pumping efficiencies due to the prevention of Er clusters and the nonradiative cross-relaxation effect between Yb and Er ions, and (2) the active double-clad fiber possibly provides ultralarge gain, and can support extracting the greater part of the intracavity laser as output power, therefore obtaining large pulse energy.

In this paper, by combining a mono-layer CVD graphene SA and a section of EYDCF as the gain medium, we successfully achieve a wavelength-tunable, large-energy, all-fiber passively $Q$-switched laser. By properly arranging the cavity-component positions (e.g., output coupler, graphene SA), the $Q$-switching operation can generate large pulse energy up to 1.05 μJ, minimum pulse duration of 2.6 μs, and tunable lasing wavelength of 1530.97–1546.92 nm. To the best of our knowledge, this is the largest single pulse energy in graphene-based passively $Q$-switched all-fiber lasers.

2. Mono-Layer CVD Graphene and Experimental Setup

The mono-layer graphene was grown by the CVD technique as described in the supporting material of Ref. [30]. By using a mixture of methane and hydrogen with temperature up to 1000°C, graphene films were grown on copper foils. Then graphene films were removed from the Cu foils by coating polymethyl methacrylate (PMMA) and dissolving Cu in an aqueous solution of Ammorium Persulphate. Finally, the PMMA/graphene was transferred onto the facet of a fiber ferrule. The PMMA layer served as a support preventing the graphene film from collapsing during copper etching and damaging during fiber ferrule connecting. The inset of Fig. 1 is a photograph of the fabricated PMMA/graphene fiber ferrule. One can easily see that the PMMA/graphene film has been effectively covered onto the fiber ferrule. After being transferred, the PMMA/graphene was detected by Raman spectroscopy, and the typical Raman spectrum of the graphene by subtracting the Raman peaks of the PMMA is shown in Fig. 1. The D peak is very weak, indicating good quality of our graphene. The intensity of the 2D band at the center wavenumber of $2695.1{\mathrm{cm}}^{-1}$ is much higher than that of the G band at $1586.4{\mathrm{cm}}^{-1}$. The intensity ratio (2D/G) is more than 2, and the FWHM of the 2D band is about $30{\mathrm{cm}}^{-1}$, indicating the mono-layer of our graphene [31,32].

 

Fig. 1. Raman spectrum of graphene with photograph of the PMMA/graphene fiber ferrule.

Download Full Size | PDF

In our experiment, the PMMA/graphene fiber ferrule was further connected with another new fiber ferrule to construct a fiber-compatible graphene SA. Then, the graphene SA was incorporated into a laser-diode-pumped EYDCF laser cavity. Figure 2 shows the schematic of the wavelength-tunable, all-fiber graphene-based passively $Q$-switched Er:Yb laser. A piece of 1.2 m EYDCF (Nufern, SM-EYDF-7/130) as gain medium was pumped by a 975 nm LD through a $975/1550\mathrm{nm}$ fiber combiner. A circulator was used to force the unidirectional light propagation and protect the graphene from the perturbance of the counterclockwise ASE light. A $50:50$ optical coupler was used as a broadband fiber loop mirror to reflect the laser. A fiber Fabry–Perot tunable filter (Micron Optics Inc. Serial NO. 10171; tuning range of 140 nm; insertion loss of $<1\mathrm{dB}$) was inserted into the cavity to tune the lasing wavelength. A $20:80$ fiber coupler was used to extract the 80% intracavity light as the laser output. The cavity length is about 10.2 m with total loss of $\sim 5\mathrm{dB}$. We deliberately designed the extraction of the large-part (80%) intracavity laser with the following purposes: (1) reducing the light power injected into the graphene SA and protecting it from thermal damage and oversaturablity under moderate pumping, (2) obtaining a relatively large output power (corresponding to large pulse energy), and (3) benefiting from the use of large-gain EYDCF, the pump threshold might still not be very high. The optical spectrum, the radio-frequency (RF) spectrum, the $Q$-switched pulse trains, and the average output power were monitored and measured by an optical spectrum analyzer (OSA) and a RF spectrum analyzer together with a photodetector, a 100 MHz oscilloscope, and a power meter, respectively.

 

Fig. 2. Schematic of the EYDCF $Q$-switched fiber laser.

Download Full Size | PDF

3. Results and Discussion

In the experiment, continuous-wave operation occurred at an incident pump power of 161.8 mW with a central wavelength of 1535.05 nm. Once the pump power exceeded 170.9 mW, a stable $Q$-switched pulse was observed. By continuously tuning the transmission peak of the fiber Fabry–Perot filter, the central wavelength of the laser tuned correspondingly and the stable $Q$-switched pulse trains on the oscilloscope screen remained. At a pump power of 582.4 mW, the central wavelength could be continuously tuned from 1530.97 to 1546.92 nm, covering a wavelength range of $\sim 16\mathrm{nm}$, as shown in Fig. 3. At four different operating wavelengths of 1546.92, 1541.35, 1535.52, and 1530.97 nm, we respectively recorded the typical pulse trains (Fig. 4). All of them were stable with a pulse intensity fluctuation of less than 5%. However, one can obviously see that, at the same pump power of 582.4 mW, the pulse repetition rate was changing at different wavelengths.

 

Fig. 3. Spectra of wavelength-tunable $Q$-switched operation.

Download Full Size | PDF

 

Fig. 4. Output pulse train at different wavelengths: (a) 1546.92 nm, (b) 1541.35 nm, (c) 1535.52 nm, and (d) 1530.97 nm.

Download Full Size | PDF

To further investigate the variation of the $Q$-switching operation during the wavelength-tuning process, we measured the output power and pulse repetition rate as a function of the operation wavelength, as shown in Fig. 5. Without the fiber Fabry–Perot filter, the laser exhibited $Q$-switching at $\sim 1535\mathrm{nm}$. After inserting the filter, the max output power of 24.8 mW came out at 1535.05 nm. When the $Q$-switching wavelength shifted toward either a longer or a shorter one, the output power decreased gradually. While tuning the lasing wavelength from 1530.92 to 1546.97 nm, the average output power could vary from 2.1 to 24.8 mW. The change of the average output power with different wavelengths should be attributed to the dependences of the EYDCF’s gain spectrum and the cavity loss. As shown in Fig. 5, the variation of the $Q$-switched repetition rates with the operation wavelengths is similar to that of the output power. This is easily explained as follows. With the larger output power, the intracavity laser is stronger, and the bleaching of graphene SA is faster under a faster population inversion/depletion, leading to larger repetition rates. At the same pump power of 582.4 mW, the repetition rate can vary from 7.5 to 25.5 kHz with the different operation wavelength.

 

Fig. 5. Output average power and pulse energy at different wavelengths within tuning range.

Download Full Size | PDF

Then, we concentrated on the $Q$-switched output performance at the lasing wavelength of 1535.05 nm (with the filter). Figure 6(a) shows the optical spectrum of $Q$-switched operation at the pump power of 453.38 mW, with an inset of a $Q$-switched pulse train of 21.19 kHz. The $Q$-switched pulse train was stable with low amplitude fluctuation. Confirmed by Fig. 6(b), the signal-to-noise ratio (SNR) of fundamental frequency at 21.19 kHz is about 50 dB, and the tenth harmonic is still larger than 33 dB from the inset.

 

Fig. 6. (a) Output spectrum and $Q$-switched pulse train (inset). (b) Radio-frequency optical spectrum at the fundamental frequency and the wideband RF spectrum (insert).

Download Full Size | PDF

The $Q$-switched regime remained stable with the pump power changed from 170.9 to 582.4 mW. When the pump power exceeded 582.4 mW, the $Q$-switched pulses became unstable and even disappeared. However, the $Q$-switching could be rebuilt once the pump power is less than 582.4 mW again, which indicates no damage of our graphene. It can be attributed to the thermal effect affecting the graphene, although the damage threshold does not exceed. We measured or calculated the output power, the repetition rate, the single pulse energy, and the pulsewidth as a function of the pump power. As shown in Fig. 7, when the pump power increases gradually, the output power, the repetition rate, and the single pulse energy increase most linearly, but the pulsewidth decreases monotonously. When the pump power exceeds 500 mW, the pulsewidth and repetition rate become slow to change and the drop in pulsewidth levels even stops; then the graphene SA almost saturates. The repetition rate could be continuously tuned from 7.07 to 24.48 kHz. When the laser was pumped with the maximum available launched power of 582.4 mW, the average output power was measured to be 25.6 mW and the pulse duration was 2.6 μs. The single pulse energy was calculated to be 1.05 μJ. To the best of our knowledge, this is the largest pulse energy of the graphene-based $Q$-switched wavelength-tunable all-fiber lasers so far. The large pulse energy is mainly attributed to the following: (1) the large-enough gain of the laser-diode-pumped EYDCF, (2) the power output of 80%, and (3) the lowest-power position of graphene in the laser cavity.

 

Fig. 7. (a) Pulse repetition rate and average output power. (b) Single pulse energy and pulsewidth as a function of the launched pump power.

Download Full Size | PDF

4. Conclusion

In summary, we have demonstrated a mono-layer CVD graphene-based large-energy, wavelength-tunable, all-fiber EYDCF laser. The lasing wavelength can be tuned from 1530.97 to 1546.92 nm with stable $Q$-switched operation. At the wavelength of 1535.05 nm, the pulse repetition rate can be tuned from 7.07 to 24.48 kHz. The largest average output power and the lowest pulse duration are measured to be 25.6 mW and 2.6 μs, respectively. The single pulse energy is calculated to be as large as 1.05 μJ. The large pulse energy benefits from the cavity-design optimization, the large-gain EYDCF, and the good quality of the mono-layer CVD graphene.