Publication No: 1232Search all ORC publications    

Single-sided output Sn/Er/Yb distributed feedback fiber laser

W.H.Loh, L.Dong and J.E.Caplen


A distributed feedback laser based on Sn/Er/Yb fiber is demonstrated. Unlike previous Er/Yb DFB lasers, hydrogenation is not required for grating fabrication. Single-sided output operation is obtained by directly splicing one end of the active grating to a broadband high reflecting grating, with 10 mW output power achieved for 110 mW pump power. Although hydrogenation is able to enhance the photosensitivity/grating strength further, this is actually found to degrade the laser efficiency due to increased losses.

Single frequency erbium-doped fiber lasers have been the subject of a great deal of interest for some years now, as they hold considerable promise for a variety of applications in telecommunications and sensor systems. Both distributed Bragg reflector (DBR) [1, 2] and distributed feedback (DFB) [3-5] erbium fiber lasers have been demonstrated; however, issues relating to laser efficiency and ease of fabrication are still not satisfactorily resolved. Er3+- germanosilicate fibers are often used in the realisation of dBR and DFB lasers, as their photosensitivity makes grating formation a simple task. However, the germanosilicate host limits the Er3+-concentration to relatively low levels, in order to avoid clustering problems, and hence the pump absorption is low, leading to poor laser efficiencies of <1%. Thus, more complex configurations are necessary to recover the efficiency, such as re-using the unabsorbed pump power in a MOPA configuration [1], or more recently, by an intracavity pumping scheme [6].

Alternatively, it has been demonstrated that Er3+/Yb3+-phosphosilicate fibers can make very efficient dBR lasers [7], however, the poor photosensitivity of the phosphosilicate host makes the formation of suitable gratings difficult. Although hydrogenation of these fibers can improve the photosensitivity considerably [8], it is a cumbersome procedure and also suffers from increased losses, particularly at shorter wavelengths. For fiber laser applications, the latter aspect is a serious drawback, as this implies high pump losses.

Recently, it has been shown that the use of tin as a co-dopant can improve the photosensitivity of fibers significantly, without need for hydrogenation [9, 10]. In this work, we report the realisation of 1.5 μm DFB lasers based on Sn/Er/Yb/Al-phosphosilicate fiber, where grating formation is aided by the presence of Sn, while high pump absorption is achieved via the high Yb3+ concentration which then transfers the energy to the Er3+-ions. The fiber has a numerical aperture of 0.17 and a single mode cut-off wavelength of 1270 nm. The Er3+- and Yb3+- concentrations are ~600 ppm and 2000 ppm respectively, with small signal absorptions of 20 dB/m at 1535 nm and 170 dB/m at 980 nm. A grating was written into the fiber by scanning a 100 mW 244 nm UV laser beam at a rate of 5 μm/s across a 10 cm phase mask. The fabricated grating had a bandwidth of 0.10 nm and 80% reflectivity.

Fig. 1 Lasing characteristics of (a) unhydrogenated DFB fiber laser,
(b) hydrogenated DFB fiber laser,
(c) 5 cm long hydrogenated DFB fiber laser. Inset: schematic of DFB laser.

To achieve the desired single-sided output operation from the laser, a 1.5 cm strong (»99% reflectivity, 1 nm bandwidth) grating was fabricated in a boron-GeO2 fiber with a similar numerical aperture and cut-off wavelength to the Sn/Er/Yb fiber, and the two gratings directly spliced together (Fig. 1, inset). The length of the active grating after splicing was 8.5 cm. This configuration is similar to that of many practical semiconductor DFB lasers, where a HR-coating is deposited on one facet to reduce the threshold and obtain single output operation. Like its semiconductor counterpart, it also suffers from an uncertainty in the grating phase at the termination/splice to the high reflector end, which can degrade the threshold and single mode stability. However, for fiber gratings, this can be subsequently corrected, e.g. by UV post-processing [11], so this drawback is likely to be less significant. In the event, though, we found that maximum lasing power and stable single frequency operation was obtainable by just having a slight bend in the grating, which effectively imparts a small chirp/distributed phase shift to it [12].
Fig. 1 (curve a) shows the lasing characteristic of the DFB laser. The 980 nm pump is coupled in via a fiber WDM, and the lasing output spliced to a pigtailed optical isolator to eliminate complications from external feedback. Essentially the same performance was obtained whether the laser was pumped from the front end or the back. The laser threshold is 20 mW, with an output power (measured after the isolator) of about 10 mW for 110 mW pump. (Without the high reflector at the back end, the laser could still lase, but with a very high threshold, over 100mW). Fig. 2 shows the output spectra (0.05 nm resolution) from the front and the back end of the laser.

Fig. 2 Top: Output spectrum of DFB laser from front end.
Below: Corresponding spectrum from back end.

It is seen that with 10 mW output from the front end, the rear end has negligible leakage (< 1 nW), due to the high reflectivity (-70 dB) of the back reflector. For the rear output spectrum, the spontaneous emission outside the stop-band of the HR grating is also evident on the short and long wavelength ends of the trace. Observations with a scanning Fabry-Perot interferometer showed that the laser operated stably in 2 orthogonally polarised modes (as observed in previous DFB fiber lasers [3, 4]), spaced 390 MHz apart, with no mode-hopping.

Fig. 3 Self-heterodyne rf beat spectrum of laser linewidth measurement.
White trace: Lorentzian fit with 50 kHz half-width.

The optical linewidth was measured using the delayed self-heterodyne technique with a 5 km delay line, and the rf beat spectrum is shown in Fig. 3. The operating power of the DFB laser was 10 mW, although similar linewidth values were also obtained at 1 mW output power. A reasonable fit is obtained with a Lorentzian lineshape of 50 kHz half-width (white trace), except for relative frequency deviations beyond ±0.4 MHz, where the experimental trace is limited by the instrument noise floor. The optical linewidth of the laser is thus estimated as 25 kHz. No intensity instabilities were exhibited by the laser up to the maximum output observed. The intensity noise spectrum showed the typical single peak at the relaxation oscillation frequency, which steadily increased in frequency, but decreased in amplitude, as the output power increased (Fig. 4).

Fig. 4 Plot of relaxation oscillation frequency and peak of relative intensity noise (RIN) spectrum with laser power.

Further investigations were conducted by first hydrogenating the Sn/Er/Yb fiber before grating fabrication. With hydrogenation, stronger gratings are quite easily achievable, and one of these gratings (99% reflectivity, 0.17 nm bandwidth) was spliced to a high reflecting grating in a similar configuration as before. In this case, though, a clear difference in the laser behaviour was observed depending on the direction of the pump, with higher output power obtained when pumped via the front end. However, in spite of the stronger grating, the threshold is significantly higher than in the non-hydrogenated case. Fig. 1 (curve b) shows the lasing characteristic, which has a threshold of 44 mW and a slope efficiency of only 3%. In fact, practically the same characteristic was obtained for the laser without the high reflector at the back, indicating that the DFB laser effectively consists of just the front several cm of the grating, without being able to take advantage of the high reflector at the back end. Accordingly, when the active grating was halved, to 5 cm (with the high reflector at the back), the lasing performance was observed to improve somewhat, to 20 mW threshold and 5.5% slope efficiency (Fig. 1, curve c). However, the performance is still clearly worse than for the non-hydrogenated laser.

The likely explanation for the higher threshold of the hydrogenated DFB laser, in spite of its stronger grating, is that the losses in the laser cavity have increased, causing the pump power to dissipate before it is able to traverse the full length of the grating. Indeed, a fiber cut-back pump power measurement of the active grating fiber revealed an excess fiber loss at 0.98 μm of 0.6 dB/cm (or 3 dB in 5 cm) over that of the untreated fiber. To corroborate this, we characterised the excess loss at different wavelengths, by measuring the transmission of the fiber before and after grating formation. The grating writing conditions for this set of measurements were with a 248 nm UV laser operating at 20 Hz, 0.3 J/cm2 pulse energies, and 10 minute exposure time on 15 mm lengths of fiber. Fig. 5 shows that the excess loss actually falls off quite rapidly for longer wavelengths, and was found to be negligible at 1.5 μm. It is clear, though, that such significant excess losses at the pump wavelength will impose a severe limitation on the maximum efficiency achievable in these grating-based lasers.

Fig. 5 Measured excess losses in the hydrogenated grating fiber with wavelength.

In conclusion, by the use of tin-codoping rather than the usual technique of hydrogenation to enhance the photosensitivity of an Er3+/Yb3+-phosphosilicate fiber, a DFB fiber laser has been demonstrated with 11% slope efficiency. Single sided output was achieved via the simple approach of directly splicing a high reflecting grating to the back end of the laser. The substitution of hydrogenation with tin-doping to obtain photosensitive active fibers is advantageous both from a practical fabrication viewpoint, and also as an effective means for avoiding excess fiber losses which impact adversely on the laser efficiency.

The authors would like to thank L Reekie and R I Laming for useful discussions and encouragement. This work was supported in part by the European Commission ACTS project PHOTOS. The Optoelectronics Research Centre is an Interdisciplinary Research Centre funded by the Engineering and Physical Sciences Research Council.

1. G A Ball, C E Houlton, G Hull-Allen and W W Morey, IEEE Photon. Technol. Lett., 6, 192 (1994)
2. J L Zyskind, V Mizrahi, D J DiGiovanni and J W Sulhoff, Electron. Lett., 28, 1385 (1992)
3. J T Kringlebotn, J L Archambault, L Reekie and D N Payne, Opt. Lett., 19, 2101, (1994)
4. W H Loh and R I Laming, Electron. Lett., 31, 1440 (1995)
5. M Sejka, P Varming, J Hubner and M Kristensen, Electron. Lett., 31, 1445 (1995)
6. W H Loh, B N Samson, Z E Harutjunian and R I Laming, Electron. Lett., 32, 1204 (1996)
7. J T Kringlebotn, J L Archambault, L Reekie, J E Townsend, G G Vienne and D N Payne, Electron. Lett., 30, 972 (1994)
8. J Canning, M G Sceats, H G Inglis and P Hill, Opt. Lett., 20, 2189 (1995)
9. L Dong, J L Cruz, J A Tucknott, L Reekie and D N Payne, Opt. Lett., 20, 1982 (1995)
10. L Dong, J L Cruz, L Reekie, M G Xu and D N Payne, IEEE Photon. Technol. Lett., 7, 1048 (1995)
11. J Canning and M G Sceats, Electron. Lett., 30, 1344 (1994)
12. H Hillmer, K Magari and Y Suzuki, IEEE Photon. Technol. Lett., 5, 10 (1993)

Applied Physics Letters (1996) Vol.69(15) pp.2151-2153

doi: 10.1063/1.117148

Southampton ePrint id: 78070




Copyright University of Southampton 2006