1. Introduction

Dual-wavelength laser systems are attractive for many applications ranging from interferometric sensing, real-time holography to biological imaging and sensing [1–3]. More importantly, they are a platform for optically generating signals in the microwave and terahertz region, by means of heterodyne beating techniques [4] or difference frequency generation schemes [5]. While most of these applications rely on signals from two separate laser sources, they would all benefit from dual-wavelength signals directly generated from a single laser. Besides the obvious cost benefits, considerable alignment efforts to achieve proper spatial overlap and, in the case of pulsed systems, temporal synchronization of the two separate beams can be avoided.

Despite the fact, that numerous realizations of dual-wavelength laser systems have been presented, adequate power scaling and stabilization of these sources is still challenging, especially when power-demanding applications such as the generation of tunable continuous-wave THz radiation are targeted. Although several, mostly fiber-based, dual-wavelength laser sources have been demonstrated, where different combinations of fiber Bragg gratings provide tunable dual-wavelength feedback [6–8], sufficient power-scaling of these lasers would require additional amplification stages. Since subsequent amplification requires hard-to-obtain and expensive components (i.e. high-power isolators), while also introducing additional ASE noise, a directly power scalable oscillator presents a favorable option.

Recently, high-power, multi-wavelength operation has been realized in VBG-locked fiber lasers [9, 10]. However, the relative power stability between the signals at different operating wavelengths suffered from gain competition due to homogenous broadening in the rare-earth doped fiber gain media, while the high-power output signals remained unpolarized.

In this work, we demonstrate a dual-wavelength VBG-locked fiber laser operating on two mutually orthogonal linear polarization states. The resulting dual-polarization output state not only enhances the lasers usefulness as pump source for common difference frequency conversion schemes i.e. in semiconductor crystals (GaP, GaSe, ZnTe, etc), the proposed configuration also ensures improved power stability, due to polarization hole burning (PHB) [11,12]. By locking one polarization with a transversely chirped VBG, the wavelength separation of the dual-line signal was efficiently tuned, even during laser operation, as previously demonstrated in [13].

2. Experimental setup

The setup of the dual-polarization fiber laser is depicted in Fig. 1. The main cavity configuration closely resembles the tunable narrow-linewidth Yb-doped fiber oscillator we recently demonstrated [13]. This laser delivered a high-power (>100 W), narrowband (<13 GHz) output signal tunable over 2.5 THz with excellent beam quality M²<1.2 and low noise (<0.2% RMS). By locking the laser cavity with a transversely chirped VBG (TCVBG), continuous wavelength tuning without the need for cavity realignment was possible, even during high-power operation. In order to ensure linear polarization operation, the transmission axis of an intra-cavity thin-film polarizer was oriented parallel to the slow axis of the polarization-maintaining gain fiber. Although lasing on the slow axis is preferable, since it suffers less coiling loss, laser action on the fast axis is also fully feasible. Consequently, simple enhancement of the described setup with a second VBG, that provides feedback for signals polarized parallel to the fast axis of the gain fiber, enables dual-polarization operation of the laser.

 

Fig. 1 High-power dual-polarization laser setup.

Download Full Size | PDF

For most applications power equalization between the two signal wavelengths is desirable. Apparently, equal power distribution is only achieved when the lasing thresholds for both signals are matched for identical outcoupling. The average reflectivity of the TCVBG in the original cavity design was with 91.6% considerably lower than the 99% maximum reflectivity at 1066 nm with a bandwidth of 0.22 nm. However, both reflectivities could be more closely matched by modifying the design of reference [13] so that the incident beam size on the TCVBG decreased. A smaller beam size simply leads to the beam covering a smaller amount of grating chirp, which results in increased reflectivity. In practice, we replaced the intra-cavity lens L1, which had a focal length of 12 mm, with a lens with 8 mm focal length, thereby decreasing the beam diameter on the TCVBG from 1.2 mm to 0.8 mm. Although the resulting average maximum reflectivity of 97.9%, was still lower than the 99% of the second VBG, optimal power equalization was achieved in this configuration, since the second VBG provided feedback to the signal polarized parallel to the fast axis of the gain fiber, which experiences more losses. Moreover, small differences in the output power ratio of both signals could be compensated for by slightly tilting one of the VBGs, thus introducing a small additional loss.

It should also be noted, that the modified design, rather counterintuitively, increased the linewidth of the laser signal locked by the TCVBG. This is a consequence of the reflectivity profile now more closely resembling a flat-top profile which provides less discrimination for neighboring longitudinal modes with wavelengths that are slightly off-center with respect to the main reflection peak [13]. Specifically, the linewidth thereby increased from approximately 13 GHz to roughly 20 GHz.

Finally, diode degradation of the pump laser reduced the available pump power from 140 to about 110 W, which reduced the maximum output power to 78 W. The original performance could easily be restored, but would not add any functionality to the dual-polarization operation discussed in this work.

3. Results

Initially, we evaluated the individual output power scaling of the two laser lines in order to compare with the power scaling for the dual-line operation, see Fig. 2(a). The output power vs pump power dependency for the laser only locked by the TCVBG was measured at the center of the grating at 1068.5 nm. In all cases, excellent launch efficiency of 96% for the pump laser was assumed, only reduced by the Fresnel reflection at the perpendicular cleaved fiber end facet. During the three sets of power measurements, the setup depicted in Fig. 1 remained unchanged as the identical high-power detectors PM1 and PM2 (Gentec UP55N-400W-H9) registered the output powers after the thin-film polarizer TFP2 which had a polarization extinction ratio (PER) of >23 dB. While the total output power, plotted in Fig. 2(a), for the two single line and the dual-line laser was the sum of both power detectors, we could directly confirm that the PER of both single-line signals was >19 dB.

 

Fig. 2 (a) Slope efficiency measurements for single-line (triangles, circles) and dual-line operation (squares); inset graph depicts maximum dual-line power vs wavelength separation (b) Dual-line laser spectra for different wavelength separations.

Download Full Size | PDF

As per design of the laser, no significant differences in slope efficiency between the three laser configurations were observed, as all reached more than 78 W output power at the identical launched pump power of 105 W.

Since the employed tuning method uses both gratings at normal incidence, no additional losses, otherwise present for tuning methods such as angle tuning [14], are introduced when the wavelength separation is increased. Therefore the maximum dual-line output power varied only within <1% over the investigated tuning range in accordance with measured power variations in our previous work, see inset in Fig. 2(a).

Figure 2(b) presents the tuning spectra for seven different transversal positions of the chirped grating, demonstrating the possibility of continuously tuning the wavelength separation between 30 GHz and 2 THz. Wavelength separations of less than 30 GHz could be realized, but the linewidth overlap of both signals did not allow to spectrally resolve the amount of separation. Also, even for perfect spectral overlap no detrimental effect on the stability of the two signals could be observed.

Subsequently, measurements monitoring both spectral and power stability of the signal during a period of 20 min were performed for minimum and maximum wavelength separation, see Fig. 3. The measured peak-to-peak power fluctuations in both cases were less than <3%, demonstrating good suppression of gain competition, while an optical signal- to-noise ratio of more than 50 dB was maintained during the measurement.

 

Fig. 3 Spectral (upper graphs) and power (lower graphs) stability for dual-line laser operation at (a) 30 GHz and (b) 2 THz wavelength separations.

Download Full Size | PDF

Moreover, we measured the laser spectra at one position after the thin-film polarizer, once with the half-wave plate at 0° and once at 45°. The results of these measurements, overlaid with the originally measured spectrum before the polarizer are displayed in Fig. 4 for two different wavelength separations. These measurements confirm that none of the individual laser lines PER was compromised by the dual-line laser operation of the laser, as the measured extinction ratio of >19 dB remained at the same level as previously measured for single-line laser operation.

 

Fig. 4 Polarization extinction ratio measurement for 0.14 THz (left) and 1.5 THz (right).

Download Full Size | PDF

However, since the amount of remnant orthogonal polarization is very similar in magnitude compared to the power fluctuations previously measured, we again modified the laser setup to evaluate how strongly PHB contributes to the inhibition of cross-talk between the two signals. By rearranging the two VBGs in sequence behind each other such that both signals had the same polarization state, and separating both output signals by using a third VBG with a design wavelength of 1100 nm and a maximum reflectivity of 99% under oblique incidence, we could evaluate the power stability of the dual-line laser running on a single polarization, see Fig. 5(a). For identical total laser output powers of 30 W, we compared the long-term power stability of the single- and dual-polarization lasers during a 15 min measurement period. As depicted in Fig. 5(b), we observed strong power fluctuations for the single-polarization laser where up to 80% of the total power was continuously redistributed between the two output signals. This behavior would suggest that the power fluctuations in the dual-polarization laser are mainly caused by gain competition between the first signal lasing on one wavelength and the remnant orthogonal polarization from the second signal lasing on the other wavelength. Future work on the improvement of the laser power stability will therefore focus on increasing the PER of the individual signals from each channel.

 

Fig. 5 (a) Setup for Single-polarization dual-wavelength laser, (b) power stability measurement for the dual-wavelength lasers with single-polarization and dual-polarization output.

Download Full Size | PDF

Finally, we confirmed that the beam quality was not affected by dual-line or single-line operation and remained constant at M²<1.2.

4. Discussion and conclusion

We propose a straightforward method to achieve high-power, tunable, dual-wavelength laser operation in a VBG-locked Yb-fiber oscillator. Simple tuning of the wavelength separation, even during laser operation, could be achieved by translating a transversely chirped VBG. By constraining the two laser lines to mutually orthogonal linear polarization states, stable dual-wavelength operation is promoted through PHB, which was confirmed by comparing single-polarization and dual-polarization operation of the dual-wavelength laser. The pump power limited output power of 78 W was maintained for a demonstrated wavelength splitting between 0.03 and 2 THz, while the power variation at both operating wavelengths remained <3%. As discussed in [13], the tuning range of the presented laser could be easily extended to encompass larger parts of the Ytterbium gain-bandwidth. By using a different properly designed transversely chirped VBG, wavelength splitting up to 20 THz should be feasible.

Although further power scaling is readily available, the demonstrated dual-wavelength laser should already provide sufficient power to generate cw THz radiation in the milliwatt regime. For example, following the numerical estimations in [5], a THz-source based on difference frequency generation in a 4 mm long GaP crystal using the presented laser as pump source, should potentially deliver 1 mW at 2 THz, using a Boyd-Kleinmann focusing parameter of 1.