# Fileset [jjap_ffs.pdf](https://mdr.nims.go.jp/filesets/b695a3e2-bd05-4b70-9cb1-ffcfbb942e96/download) ## Creator [TODOROKI, Shin-ichi](https://orcid.org/0000-0003-3986-1900) ## Rights ## Other metadata [In-Situ Observation of Fiber-Fuse Propagation](https://mdr.nims.go.jp/datasets/4a031804-1613-4eb5-8a3f-83d8633da21f) ## Fulltext Typeset with jjap2.cls Brief CommunicationIn-Situ Observation of Fiber-Fuse PropagationShin-ichi TODOROKI∗Advanced Materials Laboratory, National Institute for Materials Science, Namiki 1-1, Tsukuba, Ibaraki305-0044, JapanUltrahigh-speed photographs of fiber-fuse propagation in single-mode silica fibers revealed a new factor, otherthan pumping laser power that affects the morphology of damage sites: the length of the running optical dis-charge. Nine W pumping of 1480 nm light generated a discharge of∼130µm in half-width moving rapidly andleaving discrete voids along the fiber core, whereas 2-W-pumped discharge was∼27 µm long, moved slowlyand left nearly continuous thin voids. This is because 9-W-pumped discharge takes a longer time to travel itshalf-width, about 5/3 times longer than that of the 2-W-pumped discharge, which provides longer interactiontime for modifying the fiber core.KEYWORDS: optical fiber, fiber fuse, ultrahigh-speed photographyThe fiber-fuse effect has been a familiar phenomenon since the late 1980s,1,2) which is initiatedby the local heating of silica glass optical fiber to generate an optical discharge moving along the fiberto the light source, resulting in catastrophic destruction of the core region. Although it has attractedthe interest of many researchers, its mechanism has not yet been fully elucidated. This is due to thedifficulty in observing the rapidly moving bright discharge, which has a speed of about 1 m/s. Thus,the previous discussions have been limited to the pumping power, the speed of propagation and themorphology of generated damage.3)Here I propose a new parameter for describing this phenomenon, the length of the running opticaldischarge, which is obtained by ultrahigh-speed photography. Its propagation was observed undertwo conditions (pumping laser power: 2W and 9W) and the relationship between the length and theshape of generated damage is discussed. (This paper is based on a post-deadline paper presented atthe European Conference of Optical Communication held in Stockholm, Sweden, on September 9th2004,4) with some supplementary results and discussions.)Figure 1 shows the experimental setup used in this study. One end of a single-mode silica glassoptical fiber (SMF-28, Corning, core diameter: 9µm) was connected to a Raman fiber laser (PYL-10-1480, IPG Laser, 1.48µm , 10 W max.) via an optical isolator. The other end was stripped off andspliced to another five-meter-long fiber. The spliced section was mounted on a fiber holder equippedwith a ultrahigh-speed CCD (Charge Coupled Device) camera (ultima APX, Photron, monochromeversion, denoted by A in Fig. 1) and a CCD camera (B) connected to a video recorder. The end ofthe second fiber was mounted on another fiber holder to another CCD camera (C) connected with therecorder via a channel selector. Each camera was fitted with a zoom lens of appropriate magnification.∗E-mail address: TODOROKI.Shin-ichi@nims.go.jp1/11Jpn. J. Appl. Phys. Brief CommunicationIn order to initiate a fiber fuse, the end of the second fiber was inserted into a glass ferrule witha small amount of cobalt oxide powder. Next, laser light was introduced. Ignition and passage of thefiber fuse were recorded through cameras C and B, respectively, in order to determine the time for 5 mpropagation. A captured video image (speed: 30 frames per second) of an ignition is shown in Fig. 1.A set of neutral density filters was placed between the fiber and camera A.Two types of fiber fuse were demonstrated: one pumped by the maximum power of the fiberlaser (10 W) and the other by about one fifth of the maximum power. Since the fiber-fuse ignitionoften failed when the pumping power was below the maximum, the latter condition was realized bydecreasing the power during the run in the five-meter-long fiber after an ignition with the maximumpower. Considering the total loss of the light path and the decreasing speed of the pumping power,the power supplied to the fuse around the spliced section was about 9 W (14 MW/cm2) and 2 W (3.1MW/cm2).The thick lines in Fig. 2 are emission spectra of the discharge measured using a multichannelmonochrometer (S-2600, Soma optics) through a fiber probe placed near the spliced section. Thewavelength range of the emission coincides with the sensitivity region of camera A, 370–790nm.Although it is generally thought that the local temperature of the optical discharge becomes severalthousand kelvins,1–3) these spectra involve little heat radiation because of the lack of emissions in therange of 800–1100nm. This is clear when these spectra are compared with the theoretical spectra ofblackbody radiation at 5400K and 4000K, shown as dashed lines in Fig. 2. This is because the lightemission from the optical discharge is dominant compared with its heat radiation.Figures 3(1-a) and 3(2-a) show ultrahigh-speed photographs of fiber fuses pumped by 2 W and 9W, respectively. The height of these views has a 32-pixel resolution and covers the fiber diameter, 125µm. Thus, 10 pixels correspond to approximately 45µm. Since the fiber acts as a cylindrical lens, weshould note that the image near the core region is nonlinearly expanded in the vertical direction. Thesame goes for the photos in Fig. 1, Fig. 4 and Fig. 5. Unfortunately, the center of the fiber-fuse imagein Fig. 3(2-a), consisting of 3x7 pixels, was off-scale. Figures 3(1-b) and 3(2-b) show the intensityprofiles along the black lines on the photos, including those for 5 successive frames taken every 10µs.For each case, the traveling distance within one frame is smaller than the width of optical discharge.Therefore, these are regarded as static images.Figure 4 shows overexposed images of the fiber fuse pumped by 9 W. Discrete scattering pointsare clearly seen immediately after the optical discharge. These scattering points are likely to be due tovoid generation. One scattering point is generated about every 20µs.The propagation speeds were calculated to be 0.45 m/s for (1) and 1.3 m/s for (2), respectively,from the time they remained in the field of vision and the length of the field. On the other hand, theaveraged speeds in the five-meter-long fiber were calculated to be 0.64 m/s and 1.1 m/s, respectively.In the former case, the difference between these values is mainly due to the change in pumping powerafter fuse ignition. The difference in the latter case of constant pumping power implies variation and2/11Jpn. J. Appl. Phys. Brief Communicationfluctuation in the propagation speed, possibly due to some variations in the conditions such as curva-ture of the waveguide and the energy balance between pumping and dissipation.Figure 5 shows optical microscope images of the damaged fibers near the observation point ofFig. 3, focused on the generated voids inside. A series of discrete voids at an interval of about 23µm are seen for (2), whereas thin and nearly continuous voids are observed for (1). In addition, anexamination over several segments in the five-meter-long fiber for (1) revealed that the voids just afterthe ignition are the same as those in Fig. 5(2), and the void interval decreases as the traveling distanceincreases, i.e., the pumping laser power decreases.Atkins et al. reported that the fuse speed and the bubble spacing decreased with reduced inputpower.3) They also commented that near the lower limit of the pumping power of fiber-fuse prop-agation (∼2MW/cm2), “the bubble tracks become less regular and eventually evolve into aperiodiccavities∼30µm long.” Thus, our results in Fig. 5 are consistent with theirs.Considering the intensity profiles of the optical discharge shown in Fig. 3(1-b) and 3(2-b), theoptical discharge with the narrow and symmetric profile (half-width:∼27µm) left thin and nearlycontinuous voids whereas the broad and asymmetric one (∼130µm) left discrete voids (see Fig. 5).The width of the latter is more than the length of six voids. In addition, their durations, i.e., the timesfor traveling their half-widths, are estimated to be 60µs and 100µs, respectively. These results aresummarized in Table I.From this table, it is clear that although the propagation speed of the 9-W-pumped optical dis-charge is faster than that of the 2-W-pumped optical discharge, its duration is also longer due to itslarger width. This means that from the standpoint of a short segment of the silica fiber (e.g. less thanhalf-width of the optical discharge), the interaction time with the optical discharge pumped by 9Wis longer than that of the optical discharge pumped by 2 W. Thus, both the pumping power and thefusing duration are related to the morphology of generated damage, which will surely prompt furtherresearch activities on this phenomenon.In summary, ultrahigh-speed photographs (105 frames per second with an exposure time of 4µs)of fiber-fuse propagation along single-mode silica fibers, pumped by about 2 W and 9 W of 1480 nmlaser light, reveal that the morphology of fiber-fuse damage is affected not only by the pumping powerbut also by the length of the running optical discharge.The author is grateful to Mr. Kazuhide Hanaka and Mr. Akira Sakamaki (Photron Ltd.) for of-fering the use of their equipment for high-speed photography and Dr. S. Inoue (National Institute forMaterials Science) for his financial support.3/11Jpn. J. Appl. Phys. Brief CommunicationReferences1) R. Kashyap and K. J. Blow: Electron. Lett.24 (1988) 47.2) D. P. Hand and P. S. J. Russell: Opt. Lett.13 [9] (1988) 767.3) R. M. Atkins, P. G. Simpkins and A. D. Yablon: Opt. Lett.28 [12] (2003) 974.4) S. Todoroki:Proc. 30th European Conf. Optical Communication Post-Deadline Papers, Stock-holm, 2004(Kista Photonics Research Center, Sweden, 2004) p. 32. (Th4.3.3).4/11Jpn. J. Appl. Phys. Brief CommunicationFiberLaserVideoRecorderControllerUltrahigh-speedcamera5mAB CND filterFig. 1. Experimental setup and a captured video image of fiber-fuse ignition taken from camera C.5/11Jpn. J. Appl. Phys. Brief CommunicationFig. 2. Emission spectra of the optical discharge measured near camera B. Their intensities are normalized soas to be equal in height. The horizontal arrow indicates the sensitivity range of ultrahigh-speed camera A.The dashed lines are theoretical spectra of blackbody radiation at 5400K and 4000K.6/11Jpn. J. Appl. Phys. Brief CommunicationFig. 3. Photographs of fiber-fuse propagation from the camera A (1-a & 2-a) and intensity profiles along theblack lines on the photos for 5 successive frames (1-b & 2-b). The intensities of the laser (λ=1480 nm)coming from the left were about 2 W (1) and 9 W (2).7/11Jpn. J. Appl. Phys. Brief CommunicationFig. 4. Ultrahigh-speed photographs of fiber-fuse propagation for the 9 W condition. The view is the sameheight as that of Fig. 3(2-a).8/11Jpn. J. Appl. Phys. Brief CommunicationFig. 5. Optical microscope images of the fiber-fuse damage. The voids on the right appear at intervals of about23µm, which is in good agreement with that for the scattering points shown in Fig. 4.9/11Jpn. J. Appl. Phys. Brief CommunicationTable I. Summary of the results.pumping power (W) 2 9Velocity (m/s) 0.45 1.3Half-width (µm) 27 130Duration (µsec) 60 100Shape of the voids discrete thin & nearly continuous10/11