The properties of backreflected light due to voids in a fiber fuse were studied using optical coherence-domain reflectometry of a damaged fiber and real-time monitoring of the electrical (RF) spectrum. Light reflected backward at the interface of a propagating fiber fuse acquired low-frequency broadband amplitude modulation, which can be detected remotely at the source end, using an RF spectrum analyzer. For the light backreflected during propagation of a fiber fuse, we derived an analytical expression that well explained the spectral features observed experimentally in the RF spectrum. Finally, a novel method that allowed us to rapidly terminate the fiber fuse propagation (in a few milliseconds) is also shown.
©2009 Optical Society of America
The catastrophic damage in optical fibers due to high intensity optical radiation, commonly known as “fiber fuse” effect is observed in single-mode optical fibers delivering laser radiation with an average power in excess of a few watts. In particular, this takes place when defects in the fiber or dust at the fiber end-face or connector interface cause localized heating [1–9]. It is also shown that fiber fuse can occur in a fiber carrying high optical power when it is bent quite tightly . Once the fiber fuse starts, the optical discharge propagates unimpeded back along the fiber towards the light source, leaving the fiber core permanently damaged and unable to transmit light.
Currently, optical fibers are being used to deliver laser radiation with an average power as high as a kilowatt from high-power fiber lasers and amplifiers operating in a single spatial mode. Fibers delivering such high power levels are highly susceptible to damage from fiber fuse propagation. Also, in future broadband long-haul transmission systems, especially those employing broadband Raman amplification, it is likely that the power carried in the transmission fiber will be of the order of 1 W or more. Hence it is crucial to develop techniques to protect the source and power delivery fibers from this catastrophic damage. A short taperd fiber section with an effectively larger mode field diameter [11,12] or special hole-assisted fiber  can prevent a fuse from propagating further. However the ability to quickly detect a propagating fuse remotely would allow the opportunity to cut off the optical power and minimize the fiber damage and protecting the laser system.
We have recently observed that light reflected backward from a propagating fiber fuse undergoes amplitude modulation, and thus monitoring of the RF spectrum of the backreflected light offers a very useful means to detect fiber fuse phenomena . In this work, we have further investigated the temporal characteristics of the backreflected light and the mechanism that causes the amplitude modulation. Measurement of a reflection profile by optical coherence-domain reflectometry (OCDR) of the damaged fiber indicated that reflection comes from a few hundred micrometers of damaged fiber near the leading edge of the propagating fiber fuse. Based on this observation, we derived an expression for the backreflected light that well explained the features observed in the electrical spectrum. Also, we constructed a simple device that can rapidly detect (in about 5 ms) onset of the fiber fuse propagation, and that gives a change in the device’s electrical output voltage (DC) that can be readily used to shut down the source and halt the propagation of the fiber fuse.
2. Distributed reflection of Optical Fiber Fuse
When the fiber fuse starts propagating towards the source, it leaves behind periodic micro voids. The periodic voids, reportedly filled with oxygen , are created as a consequence of the Rayleigh instability that is due to capillary effects in the molten silica that surrounds the vaporized fiber core . The periodic change in refractive index along the fiber due to the presence of the voids results in distributed reflection from the air–glass interface. Moreover, when new voids are formed from the leading bubble surrounded by the molten silica, the shape should change with time in a periodic manner. This should result in modulation of the Fresnel reflection at the air–glass interface, resulting in amplitude modulation of the backreflected laser radiation.
Figure 1 shows microscope images of the damage caused by the fiber fuse effect in two lengths of single-mode optical fiber (SMF-28). The photographs were taken after terminating the fiber fuse propagation by rapidly turning-off the optical source. One can see micrometer-size voids, which were periodically created from the foremost large void located at the front end. Although the launched powers were the same in both cases, differences in the shape of the leading void can be clearly seen. The incoming light experiences a large index difference at the glass–air interface as it encounters the periodic voids in the core. Because of the large index modulation along the fiber, light is expected to be reflected mostly from a short section of the fiber at the leading edge. To estimate the length of fiber that contributes a significant amount of reflection, we measured the profile of the distributed reflection using OCDR [17–20] of a single-mode fiber that had been damaged by a fiber fuse.
OCDR of the damaged core was performed using the experimental setup shown in Fig. 2. Light from a 1.55 μm superluminescent diode (3 dB bandwidth: 50 nm) was divided into two parts using a 3 dB fiber coupler. The fused fiber sample was spliced to one branch, whereas light from the other branch was collimated and sent to a reference mirror. The mirror was dithered using a PZT actuator, and the output at the fourth port was detected using a photodiode. The amplitude of the fringes observed in the photodetected signal was measured at different mirror positions. Figure 3 shows the A-scan , which is a plot of the electric field reflection coefficient at different points of the damaged core, obtained by plotting the amplitude of the fringes as a function of mirror position. From the OCDR measurements, we see a distributed reflection that comes mostly from a few hundred micrometers of the damaged fiber core near the leading edge. A large reflection was caused by the leading void, and successive reflections from the periodic voids decayed quickly with the distance from the leading edge.
However, in the case of a propagating fiber fuse, the molten region propagates at a fixed velocity v F (the fiber fuse velocity). Thus, the fiber region that causes the distributed reflection can also be understood to move at the same speed. Therefore, when observed from a frame of reference that moves towards the source with the same velocity v as the fiber fuse, one should observe a reflection profile as shown in Fig. 3. Moreover, since new voids are periodically formed, the distributed reflection pattern is also expected to change in a periodic manner. The period will be the same as the time required to form a new void, i.e. Tc = p/v F, where p is the pitch. If we divide the window into N segments (N>>1), the total reflection coefficient can be thought of as a summation from each of them. Furthermore, if we assume that the reflection coefficients from individual planes are ai, and the round-trip delay times to and from each segment with respect to an arbitrary stationary reference plane are τi, the electrical field of the backreflected light from a propagating fiber fuse due to monochromatic laser radiation can be expressed as
where the second term takes into account the presence of any back-reflection due to other sources (such as the connector interface or in-line optical components). Here, a r and t r are the reflection coefficient and round-trip delay corresponding to such spurious back-reflection (for simplicity we considered spurious reflection occurring from a single plane). The expression for the optical power of the back-reflected light becomes,
A careful inspection of Eq. (1b) reveals the following:
- Since ai(t) is a periodic function with a period of Tc, i.e. ai(t + Tc)=ai(t) the second term of the equation represents amplitude modulation of the back-reflected light with period Tc. This contributes to discrete frequency components with an interval of fc = 1/Tc in the electrical spectrum of the back-reflected light.
- The cosine function in the third term represents fast sinusoidal modulation due to beating between the Doppler-shifted light from the propagating fiber fuse and the spurious back-reflection. The beat frequency is fDOOPPLER = 2nvF/λ. This term, however, vanishes for ar = 0, i.e., if there is no back-reflection.
- The fourth term represents the contribution due to interferences among light reflected at different void locations.
3. Characterization of electrical spectrum of backreflected light
We characterized, in real time, the back-reflected light from a propagating fiber fuse in the single-mode fiber (SMF-28) by application of laser radiation. Continuous-wave light from an external-cavity laser operating at 1.55 μm was amplified using a high-power EDFA (36 dBm) and was introduced into a short length (10–20 m) of single-mode optical fiber. The fiber end was cleaved and the optical fuse was initiated by applying white paint at the cleaved fiber end. Light that was backreflected towards the source during fiber fuse propagation was detected using a photodetector (bandwidth: DC to 125 MHz). The output electrical signal from the photodiode was analyzed using an RF spectrum analyzer.
Before initiation of the fiber fuse, the RF spectrum of the back-reflected light was free of any frequency components, as shown in Fig. 4(a). However, as soon as the fiber fuse was initiated, the intensity of the electrical signal in the RF spectrum increased drastically. Figure 4(b) shows the electrical spectrum observed for the case of a fiber fuse propagating in the SMF-28 with an average power of 2.75 W. Consistent with our analytical treatment, we could clearly see the fundamental modulation frequency component, fc, as well as many of its harmonics in the spectrum. The intensity was found to be 40 dB higher than the noise floor.
In addition, as expected from Eq. (1b), a Doppler frequency component fDOOPPLER = 2nvF/λ, at 876.6 kHz, could also be observed in the RF spectrum. Using, n = 1.5 and λ = 1.55 μm, we can determine the velocity of fiber fuse propagation, vF = 0.45 m/s, which agrees with the measured fiber fuse propagation velocity.
Figure 5 shows a plot of the measured electrical frequencies as a function of the harmonic number. A linear relationship can be clearly seen. The fundamental frequency fc (= vF/p) was measured to be 31.06 kHz, yielding a pitch p of 14.5 μm, which matches perfectly with the pitch measured from microscopic observation of the damaged fiber. The presence of frequencies with harmonics up to 15 in the RF spectrum suggests that the actual formation of bubbles or voids may occur on a microsecond time scale.
Besides providing an indication of fiber fuse propagation, the electrical spectrum of the back-reflected light can thus be conveniently used to characterize the features of fiber fuse propagation, such as the propagation velocity (vF) and the pitch of the periodic defects (p), which are governed by fiber types and conditions of laser excitation, such as wavelength and power.
4. Rapid detection and termination of fiber fuse
Based on the fact that the RF spectrum of the back-reflected light changes drastically when a fiber fuse occurs, we constructed a simple device that can be used to instantly terminate the propagation of a fiber fuse and protect the fiber or the source from catastrophic damage. A schematic diagram of the device is shown in Fig. 6. The device (shown enclosed by dotted square) was located after the high-power fiber source and comprised a ~1% optical coupler that extracted a small fraction of the light back-reflected from the fiber fuse towards the source, a low-speed photodetector, a DC-blocking filter, and an electrical power sensor which yields a DC output proportional to the electrical power of the low-frequency component of the photodetector’s output.
When the fiber fuse started, we observed a rapid change in the output of the electrical power sensor. The temporal change in output voltage of the electrical power sensor is shown in Fig. 7. Before the initiation of the fiber fuse, the photodetector output was essentially a DC voltage that was proportional to the average power of the back-reflected signal. Because of the presence of the DC-blocking filter, the electrical power measured by the power sensor was thus very low, typically about −55 dBm (determined by the noise floor of the photodetector, and the power sensor), thus giving a correspondingly low voltage at the output. When the fiber fuse started to propagate, a sharp rise (with a rise time as small as 5 ms) in the output voltage of the electrical power sensor could be seen. It is noteworthy that, while the average power of the back-reflected light did not increase as the fiber fuse occurred, the electrical power of the frequency components of the photodetected signal increased significantly; thus, measuring this yielded a powerful way to identify the fiber fuse event. The length of the fiber damaged by the fiber fuse that propagated at a velocity of 0.45 m/s within this short time was only about 2 mm. When the input optical power was reduced to a value lower than the fiber fuse threshold power, the output voltage of the device again dropped to the original ‘low’ voltage state. Such a change in the output voltage state can be readily used to shut down the source and halt the propagation of the fiber fuse.
In our experiments, we used single longitudinal mode pump sources. However in transmission systems, where data may propagate bi-directionally, the effectiveness of this protection system would depend on a number of factors, such as data rate, pattern length, wavelength of the signals, the modulation format, as well as the choice of bandwidths of the photodetector and electrical power sensor of the protection system. This is the scope for further studies in the near future.
In conclusion, we have studied the temporal characteristics of light back-reflected from a propagating fiber fuse using an electrical spectrum analyzer. OCDR measurements were performed to characterize the distributed reflection profile due to the voids, from which we derived an analytical expression of the reflected light associated with a propagating fuse. We have shown that the periodic formation of voids from the molten core changed the reflection profile and hence caused an amplitude modulation, which could be observed in the electrical spectrum of the back-reflected light. We have also demonstrated a simple device that is responsive to a propagating fiber fuse, helping to rapidly turn off the optical source to halt the fuse propagation.
References and links
1. R. Kashyap and K. J. Blow, “Observation of catastrophic self-propelled self-focusing in optical fibers,” Electron. Lett. 24, 47–49 (1988). [CrossRef]
3. D. D. Davis, S. C. Mettler, and D. J. Digiovanni, “A comparative evaluation of fiber fuse models in laser-induced damage in optical materials,” Proc. SPIE 2966, 592–606 (1997).
4. Y. Shuto, S. Yanagi, S. Asakawa, M. Kobayashi, and R. Nagase, “Fiber fuse generation in single-mode fiber-optic connectors,” IEEE Photon. Technol. Lett. 16, 174–176 (2004). [CrossRef]
5. Y. Shuto, S. Yanagi, S. Asakawa, M. Kobayashi, and R. Nagase, “Fiber fuse phenomenon in triangular-profile single-mode optical fiber,” J. Lightwave Technol. 24, 846–852 (2006). [CrossRef]
6. E. M. Dianov, I. A. Bufetov, A. A. Frolov, Y. K. Chamorovsky, G. A. Ivanov, and I. L. Vorobjev, “Fiber fuse effect in microstructured fibers,” IEEE Photon. Technol. Lett. 16, 180–181 (2004). [CrossRef]
7. E. M. Dianov, I. A. Bufetov, A. A. Frolov, V. M. Mashinsky, V. G. Plotnichenko, M. F. Churbanov, and G. E. Snopatin, “Catastrophic destruction of fluoride and chalcogenide fibers,” Electron. Lett. 38, 783–784, 2002. [CrossRef]
8. M. M. Lee, J. M. Roth, T. G. Ulmer, and C. V. Cryan, “The fiber fuse phenomenon in polarization-maintaining fibers at 1.55 μm,” in Conference on Lasers and Electro-Optics, Technical Digest (Optical Society of America, 2006), paper JWB66.
9. S.-I. Todoroki, “In-situ observation of fiber-fuse propagation,” Jpn. J. Appl. Phys. 44, 4022–4024 (2005). [CrossRef]
10. R. M. Percival, E. S. R. Sikora, and R. Wyatt, “Catastrophic damage and accelerated aging in bent fibers caused by high optical powers,” Electron. Lett. 36, 414–416 (2000). [CrossRef]
11. D. P. Hand and T. A. Birks, “Single mode tapers as fiber fuse damage circuit breakers,” Electron. Lett. 25, 33–34 (1989). [CrossRef]
12. R. Wyatt, R. M. Percival, and R. Kashyap, “Optical communication system and method of protecting an optical route,” U.S.A. Patent 7,162161 (2007).
13. K. Takenaga, S. Omori, R. Goto, S. Tanigawa, S. Matsuo, and K. Himeno, “Evaluation of high-power endurance of bend-insensitive fibers,” in Optical Fiber Communication Conference, Technical Digest (Optical Society of America, 2008), paper JWA11.
14. K. S. Abedin and T. Morioka, “Remote detection of fiber fuse propagation in optical fibers,” in Optical Fiber Communication Conference, Technical Digest (Optical Society of America, 2009), paper OThD5.
15. R. Kashyap, “High average power effects in optical fiber and devices,” Proc. SPIE 4940, 108–117 (2003). [CrossRef]
18. D. Huang, E. A. Swanson, C. P. Lin, J. S. Schuman, W. G. Stinson, W. Chang, M. R. Lee, T. Flotte, K. Gregory, C. A. Puliafito, and J. G. Fujimoto, “Optical coherence tomography,” Science 254, 1178–1181 (1991). [CrossRef] [PubMed]
19. J. M. Smith, “Optical coherence tomography: A review,” IEEE J. Sel. Top. Quantum Electron. 5, 1205–1215 (1999). [CrossRef]