An optical discharge running through a single-mode silica glass fiber during fiber fuse was observed and the front part of the generated damage was examined. Their pump power dependences were investigated using a 1.48 μm laser light at powers ranging from 1.1 to 9.0 W. Periodic voids were left by an optical discharge that was in a cavity with a tail. The tail appears because the optical discharge is strongly enclosed in core region. Another mode of periodic void formation was found at near the threshold pump power for fiber fuse propagation. The optical discharge in this case also forms a transient tail during the void formation cycle.
© 2005 Optical Society of America
The fiber fuse effect was found in the late 1980s. It is initiated by the local heating of an optical fiber, which generates an optical discharge running along the fiber to a light source (~W) (see Fig. 1). This results in the catastrophic destruction of the core region, i.e. the formation of periodic and bullet-shaped voids [1, 2]. The recent increase in available laser power (>kW) has led to a practical need to terminate this phenomenon once occurred [3, 4, 5]. However, the void formation mechanism is not fully understood and has only been discussed from a theoretical point of view [6, 7] and by means of a few direct observations near bright and rapidly moving (~m/s) optical discharges [8, 9, 10, 11].
Recently, the author proposed a mechanism for this periodic void formation during a fiber fuse based on a statistical analysis of a number of micrographs showing the front part of fiber fuse damage . In this model, the optical discharge exists in a long cavity in the core region of the fiber. During the fiber fuse propagation, the running cavity sheds its tail, which shrinks to form a periodic void. Its shape becomes bullet-like as the result of the balance between the internal pressure of the optical discharge and the increasing viscosity of the surrounding glass that occurs during pinching off.
However, the model is inapplicable to the examples of non-periodic damage formation reported elsewhere [2, 6, 8, 11, 13, 14, 15]. For example, an optical discharge propagating through a single-mode silica glass fiber pumped by 2.0 W light (wavelength: 1.48 μm) leaves long and narrow non-periodic voids whereas that pumped by 9.0 W leaves periodic voids [8, 11]. Therefore, this paper discusses the morphology of the optical discharge and generated voids as a function of pump power in order to clarify the factors controlling this phenomenon. Such a discussion is useful for evaluating the potential of these voids for use as photonic structures and micro capillaries.
Two types of photography were performed; ultra-high speed videography on a running optical discharge and static optical micrography on the front part of generated damage trains.
2.1. Ultra-high speed videography
Figure 2 shows the experimental set up used for this study. One end of a commercial single-mode silica glass optical fiber (SMF-28, Corning) was connected to a Raman fiber laser (wavelength: 1.48 μm). The other end was folded and brought into contact with a metallic plate in order to initiate a fiber fuse when a laser light (≥ 7.0 W) was launched into it. In order to generate an optical discharge pumped at less than 7.0 W, the laser power was reduced immediately after the fiber fuse ignition. The optical discharge was observed in a stripped section of the fiber through an ultrahigh-speed charge coupled device (CCD) camera (ultima APX-RS monochrome version, Photron Ltd., sensitivity range: 380–790 nm) with an appropriate zoom lens. Images with a resolution of 128×16 were taken every 4 μs with a 1-μs-exposure time through neutral density (ND) filters (×16 or ×32).
2.2. Observation of fiber fuse damage
The samples were prepared by switching off the pump laser after a fiber fuse was initiated and the laser energy reached an appropriate value. It required no more than 7 μs for the emission from the optical discharge to drop to zero . This value is near the resolution of the CCD camera. Photographs of the samples, which were immersed in index matching oil, were obtained with a digital microscope.
3.1. Ultra-high speed videography
Figure 3 is an edited video showing the propagation of optical discharges pumped at various laser powers (1.5–9.0 W). Figure 4 shows images captured from the video. Since the fiber acted as a cylindrical lens, these images were expanded in the vertical direction. The lengths of the horizontal arrows shown in the left row are in proportion to the speed of the optical discharge, i.e. the distance that the discharge moves during 10 frames of video (40 μs). The velocities, v, are also plotted in Fig. 5(left) as closed circles. These snapshots clearly show that the intensity profile along the fiber axis becomes asymmetric when the pump power is more than 2.0 W. Only the discharge pumped at 9.0 W has a small separate peak on the right (see the downward arrow in Fig. 4(b-1)), whose intensity oscillates between its maximum and zero every four or five frames (16~20 μs, see Fig. 3). The widths at 10% of the hight of these intensity profiles, ∆10%, are plotted in Fig. 5(right) as closed triangles.
3.2. Observation of fiber fuse damage
Figure 6 shows typical images of the damage trains. The damage generated by a laser power of more than 2.0 W (Fig. 6(a)–(d)) consists of a long and tapered void and successive small periodic voids, whereas that generated by 1.5 W (f) consists of a small droplet-like void and successive thin voids. The damage generated by 2.0 W light (e) appears similar to that generated by 1.5 W light but sometimes short segments appear that include periodic voids. The intervals of these periodic voids, d, are plotted in Fig. 5(left) as open diamonds. The damage generated by light with a power of less than 1.5 W is described later.
From the two values plotted in Fig. 5(left), that is, the propagating speed of the optical discharge and the void interval, we can calculate the period needed for the formation of one void. This value varies from 18.7 μs (9.0 W) to 31.0 μs (2.0 W). It should be noted that the micrographs showing the periodic voids (Fig. 6(a)–(d)) are only one snapshot taken during this one void formation sequence. Other moments can be captured by preparing further samples. A number of micrographs were collected and sorted in order of increasing distance between the top of the first large void and the top of the first regular void. Figure 7 shows sorted micrographs of the damage generated by 5.0 W light. This sorting operation corresponds to a rearrangement in chronological order within the void formation cycle, since the optical discharge runs at a constant rate during this cycle .
From these micrographs, we can notice that the length of the large front void varies within the one void formation cycle. As the top of the front void moves toward the light source, i.e. to the left, a neck appears in the middle of the void (Fig. 7(b)) and moves to the right (c, d). Then the neck becomes a bridge dividing the large void (e) and the bridge moves to the right (f, g, h(=a)) to form a periodic void (b).
Then, two kinds of the void length were measured and plotted in Fig. 5(right); one is for the large front void not including the bridge plotted as thin vertical bars, and the other is for that including the bridge plotted as thick vertical bars. This figure also includes the length of the front void generated by 1.5 and 2.0 W light plotted as open squares, and the maximum diameter of the void, 2r, plotted as open circles.
The morphology of the damage generated by laser light of less than 1.5 W was completely different and very sensitive to the pumping power. More than 30 samples were prepared with a laser energy of 1.1–1.4 W. In a few cases, the optical discharge self-terminated. Thus, the lower limit of the pump power for propagation is near these values. Two typical cases of generated damage are shown in Fig. 6(g) and (h). Most of the samples look like Fig. 6(h), or hybrid shapes of Fig. 6(f) and (g) or Fig. 6(g) and (h). It was rare to find samples that looks like Fig. 6(g). They are all shown in Fig. 8 after the order had been sorted in the same way as for Fig. 7.
The first point for discussion is the relation between optical discharge and the large front void. Then, the origin of periodic void formation is examined.
4.1. Optical discharge and large front void
Although a variety of void shapes are observed as shown in Figs. 6, 7, and 8, it should be noted that these are not in-situ observations of running optical discharges. These are the frozen cavities that became visible after a sudden termination of the pump laser for less than 7 μs . Though these shapes might have changed during this short quenching period, the amount of modification is not expected to be large enough to influence such typical characteristics as necks and bridges in the middle of the large front void for the following reason. The viscosity of silica glass is known to increase steeply with decreasing temperature. Since the heated area near the core region is surrounded by a cold and thick cladding layer and polymer coating, the temperature is expected to drop immediately after the laser is switched off. Therefore, considering the period of few tens of μs needed for the formation of one void at elevated temperature, the modification in a large void during this shorter quenching period is expected to be smaller in scale than that during the formation of one void.
Then, we have to estimate the location of the optical discharge in the remaining damage before the laser termination. It was located in the first large void for the following reason. As shown in Fig. 5(right), the pump power dependence of ∆10% agrees well with that of the length of the large front void (compare vertical bars and ▲ in Fig. 5(right)). Thus, the asymmetric intensity profiles shown in Fig. 4(b-1)–(b-4) result from the emission from the discharge in the tailed cavities as shown in Fig. 6(a)–(d). Examining Fig. 5(right) in further detail, we can notice that ∆10% for I = 9.0 W is located outside the thin vertical bar whereas ∆10% for 3.5 ≥ I ≥ 7.0 is within the bar. This means the emission from the optical discharge pumped at 9.0 W comes from a longer region than the large front void not including the bridge. In other words, the emission also comes from the pinched-off void. This is related to the appearance of a discrete peak in the intensity profile of the 9.0-W-pumped optical discharge shown in Fig. 4(b-1). The oscillation cycle of this peak (16–20 μs) coincides with the period of one void formation, 18.7 μs.
Figure 5(right) also shows that the maximum diameter of the front void remains constant at 2r ~ 8.6 μm. This means that the optical discharge is strongly enclosed in core region, especially for I ≥ 3.5 W. The shape of the large front void looks like a droplet when I ≤ 2.0 W but it involves a cylindrical segment when I ≥ 3.5 W. Thus, the length of the optical discharge increases linearly with increasing pump power.
4.2. Origin of periodic void formation
The periodic void formation mechanism previously proposed by the author  is demonstrated in a video shown in Fig. 7. During the fiber fuse propagation, the optical discharge in the large front void sheds its tail (see Fig. 7(e)), which shrinks to form a periodic void. This shedding action to form a bridge compensates for the creation of a new free surface at the front end of the optical discharge. Thus, the total surface area surrounding the discharge is kept in balance by these actions. This mechanism is valid when the cavity in which optical discharge exists has a tail as shown in Fig. 6(a)–(d).
This mechanism is also valid for another condition for producing periodic voids, i.e., I ~ 1.3 W (see Fig. 6(g) and Fig. 8). In this case, a tail appears transiently during the void formation cycle (see Fig. 8(b)). After that, it is pinched off (c) to form one of the periodic voids (a).
Then, what determines whether a tail appears or not? The author believes that it is the temperature distribution of the glass surrounding the front large void. If the temperature is homogeneous, the cavity minimizes its surface rather than extending it to form a tail. Fig. 6(e) and (f) seem to fall into this category. Such homogeneity is no longer attainable, however, when the optical discharge is strongly enclosed in the core region due to high pump energy, i.e., I > 2.0 W. In this case, the temperature decreases with distance from the top of the front void. This homogeneity is also impossible when the pump energy is too small. If the temperature of the rear glass is lower than that at the front, the viscosity of the rear glass is higher than that at the front. Thus, the response of the action for reducing the surface area at the rear is slower than that for creating a new surface area at the front. Therefore, the transient tail shown in Fig. 8(b) probably appears in order to make up for the delay of surface reduction by the least plastic flow obtainable from its lower viscosity. This explanation is also applicable to I ≤ 3.5 W. The viscosity of the surrounding glass at the end of the tail is higher than that at the middle of the front void. Thus, the surface area is more easily reduced by shedding the tail than a microscopic plastic flow at the end of the tail. By contrast, when the viscosity of the surrounding glass is nearly homogeneous, such a delay in surface reduction cannot occur. Consequently, the periodic void formation during a fiber fuse occurs when the optical discharge is in a cavity with a tail.
Lastly, the formation of the dual periodic structure shown in Fig. 6(h) requires comment. Considering that the pump laser power for this structure is near the critical value for fuse propagation, this structure is probably formed as a result of the fiber fuse repeatedly stopping and restarting. In order to restart the fiber fuse after self-termination, the incident laser should be focused at one position. Such a situation has been already reported in work on the in-situ observation of fiber fuse ignition . Before the appearance of an optical discharge, a short forerunning phenomenon was observed in which a dark radiant moved slowly through the core region without leaving any void. This radiant must be the focused point of incident laser beam. A similar phenomenon can occur in the present case. Thus, the void-free segments shown in Fig. 6(h) are probably generated in the period between self-termination and re-ignition. Further investigation is needed including the in-situ observation of this propagation mode.
The pump power dependence of the shape of an optical discharge during a fiber fuse and the front part of the generated voids were investigated. Before a sudden termination of the pump laser, the optical discharge existed in the front cavity. Periodic void formation was observed when the cavity had a tail. Thus, during the fusing action, the tail is pinched off and shrinks to form one of the periodic voids. The tail exists permanently when the optical discharge is strongly enclosed in the core region as a result of a high pump power. A transient tail appears when the pump laser power is near the threshold value for fiber fuse propagation. In both cases, the tail is shed to make up for the delay in the surface reduction needed to compensate for the creation of a new surface area at the front of the discharge. The delay occurs because the surrounding glass is more viscous at the rear than at the front.
The author is grateful to Mr. Kazuhide Hanaka, Mr. Akira Sakamaki and Mr. Joji Kuwabara (Photron Ltd.) for helping with the ultrahigh-speed videography experiment, and Dr. Satoru Inoue (National Institute for Materials Science) for continuous support.
References and links
1. R. Kashyap and K. J. Blow, “Observation of catastrophic self-propelled self-focusing in optical fibres,” Electron. Lett. 24, 47–9 (1988). [CrossRef]
3. D. P. Hand and T. A. Birks, “Single-mode tapers as ‘fibre fuse’ damage circuit-breakers,” Electron. Lett. 25(1), 33–34 (1989). [CrossRef]
4. S. Yanagi, S. Asakawa, M. Kobayashi, Y. Shuto, and R. Naruse, “Fiber fuse terminator,” in The 5th Pacific Rim Conference on Lasers and Electro-Optics, vol. 1, p. 386 (2003). (W4J-(8)-6, Taipei. Taiwan, 22–26 Jul. 2003),
7. S. I. Yakovlenko, “Plasma behind the front of a damage wave and the mechanism of laser-induced production of a chain of caverns in an optical fibre,” Quantum Electron. 34(8), 765–770 (2004). [CrossRef]
8. S. Todoroki, “In-situ observation of fiber-fuse propagation,” in Proc. 30th European Conf. Optical Communication Post-deadline papers, pp. 32–33 (Kista Photonics Research Center, Stockholm, Sweden, 2004). (Th4.3.3).
9. I. A. Bufetov and E. M. Dianov, “Optical discharge in optical fibers,” Physics-Uspekhi 48(1), 91–94 (2005). [CrossRef]
10. I. A. Bufetov, A. A. Frolov, E. M. Dianov, V. E. Fortov, and V. P. Efremov, “Dynamics of fiber fuse propagation,” in OFC/NFOEC 2005 Technical Digest (Anaheim, 2005). (OThQ7).
11. S. Todoroki, “In-situ observation of fiber-fuse propagation,” Jpn. J. Appl. Phys. 44(6A), 4022–4024 (2005). [CrossRef]
14. E. M. Dianov, V. M. Mashinskii, V. A. Myzina, Y. S. Sidorin, A. M. Streltsov, and A. V. Chickolini, “Change of refractive index profile in the process of laser-induced fiber damage,” Sov. Lightwave Commun. 2, 293 (1992).
15. D. D. Davis, S. C. Mettler, and D. J. DiGiovani, “Experimental data on the fiber fuse,” in 27th Annual Boulder Damage Symposium: Laser-Induced Damage in Optical Materials: 1995, H. E. Bennett, A. H. Guenther, M. R. Kozlowski, B. E. Newnam, and M. J. Soileau, eds., vol. 2714 of SPIE Proceedings, pp. 202–210 (SPIE, 1996). (Boulder, CO, USA, 30 Oct. 1995).
16. S. Todoroki, “In-situ observation of fiber-fuse ignition,” in ICONO/LAT 2005 Technical Digest on CD-ROM (St. Petersburg, Russia, 2005). (LSK3).