Open Access
Add to BibSonomy
Abstract
An efficient method to improve the efficiency of mode-field adaptation between two mismatched fibers with arcs and hydrogen-loading is proposed and demonstrated for the first time. By hydrogen-loading the fiber of a relatively smaller core, the dopant diffusion rate and the mode transition region length were significantly increased. These enhancements contributed to an abrupt diffusion rate difference at the intersection of the fibers and an adiabatic mode transition. For the mode-field adaptation of the two fibers that have a mode-field area ratio of 7.25, the transmission loss was reduced from −3.71 to −0.24 dB by an arc duration of approximately 10 seconds.
© 2019 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
A monolithic system with all fusion-spliced fiber components is free of troublesome alignment and regular maintenance. Its nature of no internal air gap provides the advantages of negligible dissipation losses and intra-cavity Fresnel reflections between the fused components that benefit a high beam quality and a great system efficiency. Due to the maturity of arc-splice technology, a monolithic fiber system with a large variety of fiber components can be easily assembled. However, for some systems that employ large mode-field area (LMA) fibers and have low tolerance of loss, such as high power fiber lasers [1–3] and passively Q-switched fiber lasers [4,5], more sophisticated fiber splicing techniques involving fiber tapering or thermally diffused expanded core (TEC) for high-end mode field (MF) adaptation are required. The TEC method has been theoretically modeled and studied thoroughly [6–9], and applied in various devices such as MF adaptors [10,11], pump combiners [12,13], fiber couplers for laser diodes [14] and mode coupling in twin-core fibers [15]. Unlike fiber tapering, which involves a physical change in the cladding layer, a typical TEC process for MF adaptation is to heat up the fiber of a small core by an oxygen-hydrogen flame to force the dopants that define the core geometry to diffuse. After the core is expanded and a desired MF area is reached, the heated zone of the fiber is cleaved and spliced with another LMA fiber. For a passive silica fiber, the core profile is determined by the Ge dopant and the diffusion coefficient of Ge is approximately 1 × 10−15 m2/s at 1400°C [6]. At this temperature, the heating time for the effective core diameter to expand from 4 to 10 μm is more than an hour. Such a long duration of heating above the glass melting point could cause fiber distortion by its own gravity and make the heated zone too fragile and difficult for the later MF adapting processes, i.e., being cleaved and spliced with another LMA fiber.
Little attention has been paid to the arc-induced TEC method mainly because of the inherent drawback of an ultrashort arc-induced TEC region of a few hundreds of micrometers [16]. To achieve an adiabatic (lossless) MF transition in such a short TEC region, the MF adaptable ratio between the two mismatched fibers is severely restricted. In addition, the short TEC region makes the later process of cleaving impracticable. Thus, MF adapting can only be performed by adding arcs at the arc-spliced intersection of the two mismatched fibers. In such a process, both the cores of the fibers at the intersection are treated and thermally expanded with the same arc power. Despite these drawbacks, MF adapting using arcs can be achieved in only tens of seconds and is still considered the most elegant and straightforward method. Here, we report a simple method that can effectively increase the arc-induced TEC region and create different diffusion rates at the intersection between the two spliced fibers by hydrogen loading the one with a smaller core diameter. In the experiment, we demonstrated that for a large MF area ratio of 7.25 between the mismatched silica fibers, the transmission loss could be reduced to −0.24 dB in an arc duration of less than 10 seconds.
Hydrogen loading of optical fibers is a well-developed technology employed for increasing the UV photosensitivity and enhancing the writing of Fiber Bragg gratings (FBG). The physics of the enhanced UV photosensitivity by H2-loading has been thoroughly studied [17,18]. The hydrogen molecules that diffuse into Ge sites produce germanium-oxygen deficiency centers (GODC) under a high temperature. An immediate 240-nm UV bleaching of the germanium-oxygen deficiency centers giving rise to the permanent change of refractive index. Despite the long development of H2-loading technology, there is surprisingly no literature reported regarding the enhanced diffusion rates of Ge by H2-loading Ge-doped silica fibers. Based on the known theory of hydrogen loading, the mechanism of the increased Ge diffusion rate was attributed to the germanium-oxygen vacancy defects created by the arc-induced high temperature of usually up to 2000 C. These vacancy defects increased the mobility as well as the diffusion rate of the Ge dopants. The physics of the diffusion enhancement by hydrogen loading is not yet clear and requires more dedicated experiments for further clarification.
2. Experiment and analysis
To monitor the instant splice loss improved by the arc-TEC treatment and preclude the inaccuracy caused by power fluctuation and wavelength sensitivity, a precise measurement system was designed and is shown in Fig. 1. The power splitter was connected to a home-made 1030-nm CW Yb-doped fiber laser, which served as a wavelength-stable light source. The laser power was set at 3 mW. One port of the power splitter was spliced with one of the fiber samples, Fiber A. The power ratio between the two ports of the power splitter, Rref = P1/P0, was first determined to be 1.53 with a standard deviation of 0.2%. During the step-by-step arc-TEC treatment, the splice transmission between Fibers A and B was monitored by measuring the power ratio between the two output ports, Rm = P2/P0, and acquired by Tm = Rm/Rref.
Fig. 1. The experiment setup designed for measuring the transmission loss between Fibers A and B and monitoring the improvement with the arc-induced TEC treatment.
In the experiment, we chose a large-core, single-mode and single-clad fiber with model number P10/125-08 made by Liekki Inc. as Fiber B, and a small-core, single-mode fiber (SMF) Hi980 made by Corning Inc. served as Fiber A. Both the original Hi980 and the H2-loaded Hi980 were tested for comparison. In the specifications, the LMA fiber P10/125-08 has a core diameter (CD) of 10 μm and a small numerical aperture (NA) of 0.08, and fiber Hi980 has a CD of 3.5 μm and a NA of 0.21. The MF areas at 1030 nm can be calculated to be 9.3 × 10−7 cm2 (Aob, for Fiber B) and 1.28 × 10−7 cm2 (Aoa) correspondingly by Marcuse's equations [19]. The MF area ratio between them was then 7.25. The H2-loaded Hi980 fiber was prepared and loaded in a gas cylinder of pure hydrogen with a high pressure of 1700 psi for 2 weeks and then tested at approximately 30 hours after being unloaded from the gas cylinder. The fiber splicer employed here was S178 LDF, made by Fitel Inc.. Fibers A and B were spliced with a standard SMF-SMF arc program, followed by step-by-step manually added arcs at the same spliced joint without shifting and pulling applied. In S178 LDF, the arc power range is scaled from 0 to 200, and set to 100 by default. The true power is not revealed in the machine specifications. Nevertheless, independent of the splicer manufacturers, the required arc power for a standard SMF-SMF splicing should be the same and can serve as a good power reference to normalize with. Therefore in the experiment, the power of every arc step was set to 100, and then the most preferable parameter to relate the output performance was the total accumulated arc duration. The transmission after every applied arc was recorded and eventually plotted as a curve, as shown in Fig. 2. Three typical transmission curves for either one of the cases using Hi980 and the H2-loaded Hi980 spliced with P10/125-08 were measured and are shown for comparison. The duration of every arc was 750 ms by default. However, due to the slower diffusion rate of the original Hi980, the duration of each additional arc was set to be 2 seconds while the original Hi980 fiber was tested. The transmission is presented as the loss in dB (y axis) corresponding to the accumulated arc duration of the applied arcs (x axis). For the original Hi980 spliced with P10/125-08, the maximum transmission with the arc-induced TEC treatment was approximately 81.5% (i.e., −0.89 dB), which could usually be achieved with an accumulated arc duration of 27 ± 2 seconds. When the H2-loaded Hi980 was employed, the optimal transmission was increased to 94.6% (i.e., −0.24 dB) achieved with a much shorter accumulated arc duration of 9.8 seconds on average.
Fig. 2. The transmission curves for splicing the LMA fiber, P10/125-08, with the small-core fibers, Hi980 and H2-loaded Hi980, improved by a step-by-step arc-induced TEC method.
It was also found that if the splicing and the TEC treatment was executed in one long arc step instead of applying multiple-step short arcs, then the transmission of −0.24 dB could be achieved with an even shorter arc duration of approximately 8 seconds. It should be noted that there was a pre-fusion duration in every arc step that was known as the time period required for heating up a fiber from room temperature to the melting point. Additionally, it was considered the time period during which the dopants did not effectively diffuse. The pre-fusion time period was 160 ms by default in the splicer. Therefore, the shorter arc duration for achieving the optimal (i.e. −0.24 dB in 8 seconds) in one arc was attributed to the less pre-fusion time. In addition, the fusion-splicing losses between the two fibers caused by the MF area mismatch, the radial offset and the angular misalignment can be calculated by the overlap integral of the MF amplitudes in the spliced fibers [20]. By assuming perfect alignment, the transmission loss caused only by the MF area mismatch can be expressed as
where RA is the MF area ratio of the two spliced fibers. Therefore, for splicing the P10/125-08 fiber with the original Hi980 fiber with no core expansion (i.e., RAo = 7.25), the theoretical transmission loss is −3.71 dB by Eq. (1).The enhancement of the Ge diffusion rates by hydrogen loading can be quantized and approximately estimated based on the results in Fig. 2 and deducted below. The calculated variables are listed in Table 1. First, the two cases are compared at a chosen transmission loss of −1.5 dB where the loss caused by the TEC transition region slope was negligible (see Fig. 4 and the later discussion). The TEC-treated MF area ratio, RA,tec, at the loss of −1.5 dB could be calculated to be 3.35 by Eq. (1). The accumulated arc durations for achieving the transmission loss of −1.5 dB were 2.25 and 10.75 seconds for the cases of using the H2-loaded Hi980 and the original Hi980, and by subtracting the pre-fusion durations described above (i.e., 2.25–0.16 × 3 and 10.75–0.16 × 6), the effective diffusion durations were 1.77 and 9.79 seconds, correspondingly. They are symbolized as τd,h and τd,o in Table 1.
Table 1. Calculation and estimation of the enhanced arc-induced TEC method by hydrogen loading based on the results in Fig. 2.
Because the P10/125-08 and the original Hi980 have the same Ge diffusion coefficient (cm2/s), their MF area differences increased by diffusion, ΔAo, should be the same and could be calculated to be 2.13 × 10−7 cm2 for reaching the referred TEC-treated MF area ratio of 3.35 by Eq. (2),
For the case of splicing P10/125-08 with the H2-loaded Hi980, due to the shorter diffusion duration, the increased area ΔAs in the P10/125-08 was 3.85 × 10−8 cm2 calculated by ΔAo ×(τd,h /τd,o). To reach the same referred MF ratio of 3.35, the increased area in the H2-loaded Hi980, ΔAHL, should be 1.61 × 10−7 cm2 by Eq. (3), Therefore, the MF expansion rate of the H2-loaded Hi980 was estimated to be approximately 4.2 times higher than that of the original Hi980.The major advantages of using the H2-loaded fiber are the extension of the arc-induced TEC region and the difference in diffusion rates at the spliced intersection that reduces the transmission loss and shortens the processing time. To demonstrate these advantages, the images near the arc-fusion zone of the spliced fibers were captured and processed to determine the core expansions, as shown in Fig. 3. Here, the fibers Hi980 and H2-loaded Hi980 were spliced together for comparison and arc-TEC treated with various accumulated arc durations of 1.5, 3 and 6 seconds. The photographs captured by the splicer are shown in Figs. 3(a)–3(c). The core contours were calculated in image processing by recognizing the darkest points that were nearest to the bright centers of the cores, and then the core diameters and the variations were retrieved and are shown in Figs. 3(d)–3(f) correspondingly.
Fig. 3. (a)-(c) The arc-induced core expansions of the spliced fibers, Hi980 (on the left) and H2-loaded Hi980 with various arc durations of 1.5, 3, and 6 seconds. Correspondingly, the core diameters were calculated from the core contours of the photos and are shown in (d)-(f).
As shown in Fig. 3, the core diameter of the H2-loaded Hi980 fiber was expanded up to 10 μm in 6 seconds with a relatively longer transition region. The comparison with the results of the original Hi980 reveals a greatly increased Ge diffusion rate by hydrogen loading. It should be noted that the way to define a CD by the contour has a precision limitation for measuring a small CD due to the low image resolution. For instance, the measured CD of the original Hi980 was close to 5 μm instead of 3.5 μm in the official specification. Therefore, the TEC region of the original Hi980 might be a little longer than that in Figs. 3(d)–3(f). In spite of the resolution limitation, the differences in Ge diffusion rates and the expansion trends are clearly shown.
To further understand the causes of loss and the limit of arc-induced MF adapting, we let Fibers A and B be the same type of fiber, spliced and kept arcing them until great transmission losses occurred. Three fiber types, the H2-loaded Hi980, Hi980 and P10/125-08, were tested and the transmission degrading curves are shown in Fig. 4. Because of the negligible losses of MF mismatch and other misalignment factors, the later emerged loss should be solely attributed to the increased transition slope of the TEC region, herein indicated by Lslp. Based on Fig. 4, it is justified to say that for splicing P10/125-08 with Hi980, as shown in Fig. 2, the loss part Lslp should be the average of the two curves of P10/125-08 and Hi980 shown in Fig. 4. For instance, with the optimal arc duration of 27 seconds (see Fig. 2), the loss Lslp was expected to be approximately −0.23 dB (i.e., the average of 0.13 and 0.33 from the curves in Fig. 4). On the other hand, for splicing P10/125-08 with the H2-loaded Hi980, the loss Lslp with the optimal arc duration of 9.8 seconds (see Fig. 2) was only −0.03 dB and there was almost no loss from the part of P10/125-08. It was noted that the loss part Lslp of −0.03 dB was better than the experimental optimal data of −0.24 dB in Fig. 2, indicating a possible loss mechanism other than Lslp and the MF size mismatch. The cause of the loss gap of 0.21 dB was not clear and might be attributed to a possible asymmetric core expansion by one-dimensional arcing. After all, perfect MF adaptation requires the match of not only the area size but also the shape at the junction of the two mismatched fibers. That is, the transmission should be further improved if the fibers were TEC-treated with a more symmetric heating source such as a three-electrode arc splicer.
Fig. 4. The transmission loss, Lslp, degraded by arcs in steps for splicing the same-type fibers (i.e., Fibers A and B in Fig. 1 are the same type). The losses Lslp resulted exclusively from the TEC transition slopes.
3. Conclusion
We have demonstrated that MF adaptation using the arc-induced TEC method could be much improved by hydrogen-loading the fiber with a relatively smaller core. For MF adapting the mismatched fibers with a large MF area ratio of 7.25, the transmission loss was reduced from a theoretical −3.71 dB to −0.24 dB in an accumulated arc duration of 9.8 seconds. The Ge diffusion rate of the H2-loaded silica fiber was estimated to be 4.2 times higher than that of the original fiber. Due to the enhanced diffusion rate of Ge by hydrogen loading, MF adaptation between two highly mismatched fibers can be efficiently achieved in a very short arc time with the fiber shape remaining unchanged. The physics of the enhanced Ge diffusion rate was attributed to germanium-oxygen vacancy defects induced by the loaded hydrogen molecules near Ge sites and the high arc temperature. More dedicated experiments and theoretical modeling are required for further clarification of the mechanism. It should be expectable that the enhancement of MF adaptation can also be achieved using various heat sources, such as CO2 lasers and O2-H2 flames.
Funding
Industrial Technology Research Institute (ITRI).
References
1. Y. Jeong, J. K. Sahu, D. N. Payne, and J. Nilsson, “Ytterbium-doped large-core fiber laser with 1.36 kW continuous-wave output power,” Opt. Express 12(25), 6088 (2004). [CrossRef]
2. B. Wang, E. Mies, M. Minden, and A. Sanchez, “All-fiber 50 W coherently combined passive laser array,” Opt. Lett. 34(7), 863–865 (2009). [CrossRef]
3. D. J. Richardson, J. Nilsson, and W. A. Clarkson, “High power fiber lasers: current status and future perspectives,” J. Opt. Soc. Am. B 27(11), B63–B92 (2010). [CrossRef]
4. D. Jin, R. Sun, S. Wei, J. Liu, and P. Wang, “Nanosecond Yb-Doped Monolithic Dual-Cavity Laser Oscillator With Large Core Fiber,” IEEE Photonics Technol. Lett. 27(14), 1477–1480 (2015). [CrossRef]
5. T.-Y. Tsai, Y.-C. Fang, H.-M. Huang, H.-X. Tsao, and S.-T. Lin, “Saturable absorber Q- and gain-switched all-Yb3+ all-fiber laser at 976 and 1064 nm,” Opt. Express 18(23), 23523 (2010). [CrossRef]
6. K. Shiraishi, Y. Aizawa, and S. Kawakami, “Beam expanding fiber using thermal diffusion of the dopant,” J. Lightwave Technol. 8(8), 1151–1161 (1990). [CrossRef]
7. K. Shiraishi, T. Yanagi, and S. Kawakami, “Light-propagation characteristics in thermally diffused expanded core fibers,” J. Lightwave Technol. 11(10), 1584–1591 (1993). [CrossRef]
8. M. Kihara, M. Matsumoto, T. Haibara, and S. Tomita, “Characteristics of thermally expanded core fiber,” J. Lightwave Technol. 14(10), 2209–2214 (1996). [CrossRef]
9. G. S. Kliros and N. Tsironikos, “Variational analysis of propagation characteristics in thermally diffused expanded core fibers,” Optik 116(8), 365–374 (2005). [CrossRef]
10. B. Wang and E. Mies, “Review of fabrication techniques for fused fiber components for fiber lasers,” Proc. SPIE 7195, 71950A (2009). [CrossRef]
11. X. Zhou, Z. Chen, H. Chen, J. Li, and J. Hou, “Mode field adaptation between single-mode fiber and large mode area fiber by thermally expanded core technique,” Opt. Laser Technol. 47, 72–75 (2013). [CrossRef]
12. K. Zhao, X. Chang, Z. Chen, Z. Wang, and H. Jiang, “Fabrication of high-efficiency pump and signal combiner based on a thermally expanded core technique,” Opt. Laser Technol. 75, 1–5 (2015). [CrossRef]
13. J. Wu, Y. Sun, Y. Wang, T. Li, Y. Feng, and Y. Ma, “The study of the thermally expanded core technique in end-pumped (N + 1)×1 type combiner,” Proc. SPIE 9255, 92550I (2015). [CrossRef]
14. Y. Lin and S. Lin, “Thermally expanded core fiber with high numerical aperture for laser-diode coupling,” Microw. Opt. Technol. Lett. 48(5), 979–981 (2006). [CrossRef]
15. Y. Zhao, A. Zhou, X. Ouyang, Y. Ouyang, C. Zhou, and L. Yuan, “A stable twin-core-fiber-based integrated coupler fabricated by thermally diffused core technique,” J. Lightwave Technol. 35(24), 5473–5478 (2017). [CrossRef]
16. M. Ratuszek, Z. Zakrzewski, and J. Majewski, “Characteristics of Thermally Diffused Transit Areas of Single-Mode Telecommunication Fibers,” J. Lightwave Technol. 27(15), 3050–3056 (2009). [CrossRef]
17. R. M. Atkins, P. J. Lemaire, T. Erdogan, and V. Mizrahi, “Mechanisms of enhanced UV photosensitivity via hydrogen loading in germanosilicate glasses,” Electron. Lett. 29(14), 1234 (1993). [CrossRef]
18. A. Othonos, “Fiber Bragg gratings,” Rev. Sci. Instrum. 68(12), 4309–4341 (1997). [CrossRef]
19. D. Marcuse, “Loss analysis of single-mode fiber splices,” Bell Syst. Tech. J. 56(5), 703–718 (1977). [CrossRef]
20. E. G. Neumann, Single-Mode Fibers: Fundamentals (Springer-Verlag, Berlin, 1988).
