It is demonstrated that by using a multimode fiber Bragg grating, the oscillation wavelength of semiconductor lasers can be selected by adjusting the alignment between the laser diode and multimode fiber. Wavelength locking with high output power and narrow linewidth can be realized in both static and dynamic states.
©2005 Optical Society of America
Low-cost semiconductor lasers with high output power, narrow spectral width, high modulation speed and direct multimode fiber (MMF) pigtails are in high demand in optical local area networks (LANs), in which the MMFs are widely used [1–2]. Besides the LANs, MMFs also have many other potential applications in optical sensors, medical instruments and optical pumping.
External cavity semiconductor lasers (ECSLs) based on single-mode fiber Bragg gratings (SM-FBGs) have been extensively studied due to their excellent properties, such as high wavelength stability against temperature and injection current, narrow linewidths and low frequency chirp under direct modulation . However, single-mode fibers (SMFs) have small core diameters, and hence the coupling efficiency between an SMF and a semiconductor laser diode (LD) is usually very low. Though multiple reflection peaks exist in the reflection spectra of an MM-FBG [4–5], the light sources based on MM-FBGs have been investigated [6–7], taking advantages of the MMF characteristics, such as large core diameters and high coupling efficiency with an LD. In particular, an external-cavity vertical-cavity surface-emitting laser (VCSEL) with an MM-FBG was reported, where the spectral width was reduced from 0.5 to 0.1 nm, and hence the transmission capacity was improved . In our recent demonstration, the oscillation wavelength of an edge-emitting laser diode (LD) can be locked at one of the reflection peaks of an MM-FBG, where the output power was significantly increased, and the spectral width was reduced to less than 0.05 nm . However, the oscillation wavelength of laser diode (LD) can be locked at only one of the MM-FBG reflection peaks, and the wavelength locking was also dependent on the FBG temperature and injection current.
In this work, we found that the reflection peak with the highest reflectivity can be selected by changing the excitation condition of an MM-FBG, and the LD oscillation wavelength can be selected by changing the alignment between the LD and MMF. This new phenomenon may lead to new applications of the MM-FBG-based ECSLs. Furthermore, different AR-coated LDs and MM-FBGs were compared in the experiments to show wavelength locking dependence on the AR coating and FBGs.
2. Experimental results
To study the MM-FBG-based ECSLs, the reflection spectrum of an MM-FBG was investigated first. In the experiment, a tunable laser source with an SMF pigtail is used, and the light is coupled from the SMF to the MMF by using a three-axis stage with a 10-nm accuracy. By changing the offset between the SMF and MMF, it is found that the reflection peak with the highest reflectivity can be changed. The specific reflection spectra with the highest reflection peak at P1 (1525 nm), P2 (1523.5 nm), P3 (1522 nm) are depicted in Figs. 1(a)–(c), respectively. It is shown that several reflection peaks with a spacing of about 1.5 nm exist in the reflection spectra of the MM-FBG. The reflection peak P1 is formed through the lowest-order self-counter-propagation mode coupling. The higher-order self-counter-propagation modes coupling form the reflection peaks located at the shorter wavelengths. The sub-peaks, such as P12 and P23, in the spectra are formed through the coupling of adjacent modes. Unlike an SM-FBG, the reflectivities of the reflection peaks of an MM-FBG are dependent on the excitation condition . When the excitation condition of an MM-FBG is changed, the mode power distribution in the MMF is changed, and hence the reflectivities related to the different mode coupling are changed. If an MM-FBG is illuminated by an LD directly, the reflectivities of the reflection peaks could be different to those via an SMF, but the method — changing the alignment between the LD and MMF — for the selection of the highest reflection peak of the MM-FBG can be used to select the LD oscillation wavelength.
The experimental setup of the ECSL with an MM-FBG is shown in Fig. 2. One facet of the LD is AR-coated with a reflectivity of about 10-4, and the other facet is cleaved with a reflectivity of about 0.32. A tapered lens fiber is used to facilitate the coupling and reduce the fiber facet reflection. The FBG is imprinted into a graded index (GI) MMF [a 62.5-µm core diameter and a 0.27 numerical aperture (NA)]. The output power and optical spectrum are monitored at the fiber output end (angle-cleaved).
The oscillation wavelength of the LD can be locked at three different reflection peaks of the MM-FBG by adjusting the alignment between the LD and MMF through a three-axis piezo-electrical stage with a step size of 0.2 µm. The wavelength locking at P1, P2 and P3 is independent of the injection current and FBG temperature. When the oscillation wavelength is locked at P1, P2 and P3 at an injection current of 40 mA, the output spectra are shown in Figs. 3(a)–(c), respectively. We can see in Fig. 3 that the side-mode-suppression ratio (SMSR) is about 43, 43 and 36 dB for the wavelength locking at P1, P2 and P3, respectively. By contrast, when the same LD and an SM-FBG with a reflectivity of about 0.55 and a Bragg wavelength of 1520.0 nm is used, the SMSR is about 40 dB.
The specific wavelength locking regions of the ECSL with the MM-FBG in terms of alignment are shown in Fig. 4. The regions for wavelength locking at P1, P2 and P3 are represented by the areas enclosed with the black, red and blue curves, respectively. One can see in Fig. 4 that the laser has a larger locking region at P1 and smaller locking regions at P2 and P3.
The corresponding output light versus injection current (L-I) curves are given in Fig. 5. The threshold current is about 19, 21, and 23 mA for the LD with the MM-FBG at the three oscillation wavelengths of P1, P2 and P3, and about 20 mA for the same LD with an SM-FBG (R=0.55, λ B=1520.0 nm). The slope efficiency is about 0.2, 0.2 and 0.14 mW/mA for the oscillation wavelength locked at P1, P2 and P3, and about 0.068 mW/mA for the SM-FBG. Compared with the SM-FBG, the output power of the ECSL with the MM-FBG is increased by 2–3 times. However, as shown in Fig. 5, the threshold current with the MMF is almost the same as that with the SMF. This can be understood by considering two coupling efficiencies between the LD and fiber: one is from the LD to the fiber (η1), and the other is inversely from the fiber to the LD (η2). Since the NAs of the LD and fiber (MMF or SMF) are different, η1 and η2 are different. In terms of η1, the MMF has a higher coupling coefficient with the LD than the SMF since the MMF has a larger core diameter and a higher NA. However, in terms of η2, the SMF has a higher coupling efficiency than the MMF. The threshold current is dependent on the external-cavity feedback determined by η 1·η 2·RFBG. As a result, though η1 is higher for the MMF, the feedback may not be more than that of the SMF. Compared with the SMF, the threshold current in the case of the MMF may not be effectively reduced. Similar experimental results were shown in . The advantages of using an MM-FBG are a higher slope efficiency and a higher output power. This is because an MMF can extract more power than an SMF due to a higher η1.
The LD was then directly modulated using an S-parameter vector network analyzer (Agilent 8719C). The output spectra and temporal waveforms were monitored by using an optical spectrum analyzer (Ando 6317) and a wide-bandwidth oscilloscope (Agilent 86100A), respectively. In this experiment, the oscillation wavelength can be stably locked at P1, P2 and P3 in the dynamic state. The frequency response curves at a bias current of 30 mA and a peak-to-peak modulation amplitude of 10 mA are shown in Fig. 6 for the oscillation wavelength locked at P1, P2 and P3, respectively, at which the modulation response curves show similar performance. Also, we can see in Fig. 6 that there are several peaks in all frequency response curves with about a 1.3-GHz spacing, corresponding to a fundamental cavity resonance frequency denoted by f=c/(2×L), where L is the optical cavity length including the external cavity length and LD length. In the experiments, the external cavity length is about 8 cm, and the LD length is about 300 µm, hence the optical cavity length is L=8 cm×1.45+300 µm×3.5=11.7 cm, corresponding to a 1.3-GHz resonance frequency. It is worth noting that this resonant enhancement in the modulation response can enable an ordinary laser chip to be modulated at a frequency higher than its intrinsic modulation bandwidth. Similar experimental results have been reported in [8–10].
The ECSLs composed of the same LD structures but with different AR coatings (R=10-4, 10-3 and 0.02) and the same MM-FBG were investigated. It is found that by using the LD with an AR coating of R=0.02, the oscillation wavelength can be locked only at P1; with an AR coating of 10-3, two oscillation wavelengths (P1 and P2) can be locked; with an AR coating of 10-4, three oscillation wavelengths (P1, P2 and P3) can be locked. Meanwhile, the stability of wavelength locking with these different AR-coated LDs was also examined. For R=0.02, the wavelength locking is dependent on the FBG temperature and injection current, as reported in . For R=10-3 and 10-4, the wavelength locking is nearly independent of the FBG temperature and injection current. It is also found that the number of lockable wavelengths and the locking stability are dependent on the residual reflectivity of AR coating. With a lower- reflectivity AR coating, the effects of the intra-cavity modes are suppressed, and hence more wavelengths can be locked, and the wavelength locking is more stable. A theoretical investigation of the AR coating effects on wavelength locking with the MM-FBG will be detailed elsewhere.
For the LD with an AR coating of 10-4, the ECSL with different MM-FBGs was also compared. These MM-FBGs were fabricated using the same phase mask but with different UV exposure doses. The more the exposure, the stronger the grating. Upon the same excitation condition, a stronger grating exhibits more reflection peaks, and the reflectivities of its reflection peaks are correspondingly higher. Wavelength locking at P4, P5, P6 and P7 can also be realized by using a stronger grating and increasing the offset between the LD and MMF, as shown in Figs. 7(a)–(d), respectively. Unlike the locking regions for P1–P3, the regions for the wavelength locking at P4–P7 are smaller, because of the weaker feedbacks. However, the locking regions for P4–P7 are expected to be enlarged by increasing the FBG reflectivities. Since the alignment required for the wavelength locking at the same reflection peak is different for different MM-FBGs, how to fabricate an MM-FBG with a required refractive index change profile to obtain a larger alignment tolerance for a desired oscillation wavelength needs to be further studied.
The reflection peak corresponding to the highest reflectivity of an MM-FBG can be selected by changing the excitation condition. This technique is then used to lock LDs at different reflection peaks of the MM-FBGs by changing the alignment between them. Stable wavelength locking at the multiple Bragg wavelengths of MM-FBGs under the static and dynamic conditions is demonstrated. By comparing the wavelength locking using the same LDs with different AR coatings, it is found that the wavelength locking is more stable when using a lower-reflectivity AR coating.
This work is supported in part by the Ontario Photonics Consortium (OPC), the Photonics Research Ontario (PRO), the Natural Sciences and Engineering Council of Canada (NSERC), and the Canada Foundation for Innovation (CFI) under the New Opportunities program. The authors would also like to thank Oki Inc. for providing the AR-coated laser diodes, and Dr. X. -Z. Chen, Dr. X. Li, and Dr. D. T. Cassidy for lending the measurement instruments.
1. J. B. Schlager, M. J. Hackert, P. Pepeljugoski, and J. Gwinn, “Measurements for enhanced bandwidth performance over 62.5-µm multimode fiber in short-wavelength local area networks,” J. Lightwave Technol. 21, 1276–1285 (2003). [CrossRef]
2. L. B. Aronson, B. E. Lemoff, L. A. Buckman, and D. W. Dolfi, “Low-cost multimode WDM for local area networks up to 10 Gb/s,” IEEE Photon. Technol. Lett. 14, 1489–1491 (1998). [CrossRef]
3. H. Bissessur, C. Caraglia, B. Thedrez, J. -M. Rainsant, and I. Riant, “Wavelength-versatile external fiber grating lasers for 2.5-Gb/s WDM networks,” IEEE Photon. Technol. Lett. 11, 1304–1306 (1999). [CrossRef]
4. K. Hwanser, K. F. Voss, and A. D. Kersey, “Novel fiber devices and sensors based on multimode fiber Bragg gratings,” Proc. SPIE 2360, 265–268 (1994). [CrossRef]
5. T. Mizunami, T. V. Djambova, T. Niiho, and S. Gupta, “Bragg grating in multimode and few-mode optical fibers,” J. Lightwave Technol. 18, 230–235 (2000). [CrossRef]
6. T. Mizunami, T. Hamada, and T. Yamamoto, “External-fiber-grating vertical-cavity surface-emitting lasers,” IEEE Photon. Technol. Lett. 12, 1558–1560 (2000). [CrossRef]
7. H. -G. Yu, C. -Q. Xu, Y. Wang, J. Wojcik, Z. -L. Peng, and P. Mascher, “External-cavity semiconductor laser with Bragg grating in multimode fiber,” IEEE Photon.Technol. Lett. 16, 2341–2343 (2004). [CrossRef]
8. K. Y. Lau and A. Yariv, “Direct modulation and active mode-locking of ultrahigh speed GaAlAs lasers at frequencies up to 18 GHz,” Appl. Phys. Lett. 46, 326–328 (1985). [CrossRef]
9. R. Nagarajan, S. Levy, A. Mar, and J. E. Bowers, “Resonantly enhanced semiconductor lasers for efficient transmission of millimeter wave modulated light,” IEEE Photon. Tech. Lett. 5, 4–6 (1993). [CrossRef]
10. Z. Ahmed and R. S. Tucker, “Small-signal IM response of grating-terminated external cavity semiconductor lasers,” IEEE J. Sel. Quantum Electron. 1, 505–515 (1995). [CrossRef]