D-shaped fibers have attracted widespread attention due to their special structure and excellent performances. However, the fabrication of D-shaped fiber faces many challenges. Mechanical polishing and chemical etching are two traditional methods for D-shaped fiber fabrication, in which micro-crack or contamination is unavoidable. In this paper, we report an efficient method to produce D-shaped fiber with a pulsed CO2 laser. The effects of processing parameters on processing efficiency and surface quality are investigated experimentally, and the processing defects such as melting debris and oscillation are also studied. The optimized processing parameters are provided.
© 2021 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
In recent years, D-shaped fiber has been used in a variety of applications, such as fiber filter , optical fiber amplifier , fiber couplers , polarizer , and fiber sensors . D-shaped fiber is an optical fiber with the cladding layer partly removed. The brittleness and small size are the two main difficulties in the processing of D-shaped fiber.
Generally, mechanical polishing and chemical etching are two conventional methods to remove cladding. In the mechanical polishing, a fiber is glued by epoxy into a cutting curved V-groove of glass or a silicon substrate , then the cladding is polished away a limited length of several millimeters . V-shaped groove with different sizes must be made precisely to satisfy fiber diameter and polishing depth. This brings high costs and low productivity. Another mechanical polishing technique is the wheel polishing technique , which is more efficient by using a grinding wheel. Long polishing distance can be realized flexibly. This method begins with the coarse abrasives to remove a large part of cladding and then followed by a fine abrasive to reduce scratches. Using various mesh sandpapers lead to high cost. In addition, defects or micro-cracks are unavoidable in both two mechanical polishing methods .
Chemical etching uses a hydrofluoric (HF) solution to remove part of the cladding. The etching rate is hard to control, because the etching rate is affected by the HF solution concentration  and temperature . While the reaction products including water , which will change the concentration of the HF solution. The reaction products such as bubbles and sols  adhere to the fiber surface, therefore slows down the reaction between HF solution and fiber, which affecting the etching rate. This leads to a rough surface  and makes the fiber brittle. The uncontrollable etching rate limits the application of this method in the precise processing of optical fiber.
For the past few years, great interests have been focused on laser processing because of its high efficiency and free of contamination. Laser processing is a non-contact processing method, which has many advantages, such as high flexibility and ultra-smooth processing surface . For these reasons, the laser is used to fabricate optical devices has been reported in many researches. Boyd utilizes the CO2 laser to cut the fiber , and Grenier uses a femtosecond laser to write waveguide into the cladding of the fiber . However, D-shaped fiber fabricated by pulsed CO2 laser has not been reported as far as we know.
In this paper, we propose to manufacture D-shaped fiber using a pulsed CO2 laser, aiming to achieve higher efficiency and quality. The effects of processing parameters on the processing efficiency and processing quality of D-shaped fiber are investigated with a home-made device. This non-contact processing method shows great potential to produce high-quality D-shaped fiber with a smooth surface and precise size.
2. Experimental material and method
The optical fibers were processed by a CO2 pulsed laser system, as shown in Fig. 1(a). The CO2 laser with 600 W maximum peak power, 200kHz maximum repetition rate, 37.5% duty cycle and 9.3 µm wavelength (fused silica has a high absorption efficiency for this wavelength ) are delivered by a SYNRAD p150 laser source. The laser pulses are focused by a ZnSe lens with a 75 mm focal length and tangentially incident to the fiber surface, resulting in a spot with 132 µm waist diameter. Besides, the experiments were conducted under room temperature (25 °C).
The optical fiber (Dcore/cladding=14/250 µm) with protective plastic coating layer partly removed (length=5cm) was held in the rotary stage which can linearly translate in three orthogonal axes and kept straight by a weight of 100 g. Firstly, the laser focal plane was positioned at the top surface of the fiber. Then the fiber was moved back and forth along its axis, in the meantime feed it up slowly, as shown in Fig. 1(b). The processed fiber includes three areas, the transition zone on both sides and the flat zone in middle, as shown in Fig. 1(b). The fibers were processed with different processing parameters including scanning speed v, pulse width w, feed distance d and number of scan n. Other laser processing parameters are list in Table 1.
To study the residual cladding thickness (RCT) (defined as the minimum cladding thickness through the core of the D-shaped section, as shown in Fig. 1(b)) and the roughness of the D-shaped fiber, a Combiner Manufacturing System (CMS from 3sae Technologies, Inc.) and white light interferometer (ZYGO NewView 9000) was employed. In order to observe the sectional profiles, an optical microscope (SCIENSCOPE NZ-PK10-E1) was used.
3. Experimental results
The surface roughness Sa, the variance S of residual cladding thickness (RCT) in the flat zone, and material removal rate (MRR) are estimated to evaluate the surface quality and the processing efficiency. The MRR was defined as MRR = V/t, where V is the volume of the material of the flat area removed and t is the time that taken to ablate material with a volume of V, expressed by following:
3.1. Pulse width
To investigate the influence of pulse width on the surface quality and efficiency, a series of pulse widths are implemented in experiments. The fiber was processed with a scanning speed of 5 mm/s, the number of scan 2, repetition rate 0.1 kHz (kept as constant to avoid fiber deformation based on our prior experiments) and the average power for the pulse width of 10 µs,15 µs, 20 µs, 25 µs, 30 µs and 35 µs are 0.634 W, 0.957 W, 1.279 W, 1.616 W, 1.953 W and 2.292 W. It can be seen from Fig. 2(a) that the MRR increases with the increase of the pulse width. In addition, the roughness shows a gentle increase and a sharp increase after 25 µs. This is because the molten debris deposited on the surface, which is shown in Fig. 2(c). The formation of molten debris is related to the pulse energy, which will be discussed in Part 4. Figure 2(b) shows that S increases with the increases of the pulse width, which means that the vibration of the processed surface becomes more intense as the pulse width increases. It can also be seen from Fig. 2(c) that as the pulse width increases, the distance between the peaks and valleys is increasing, which is the result of the glass melts and flows to form a crater due to the long pulse width. The mass loss of the glass material is accomplished by evaporative ejection, the vapor pressure is generated and reaches 1 bar  during the evaporation process, then pushes the melt outward to create a crater rim . The rim solidified into a glass wall, which increasing the distance between the peaks and valleys of the ablation zone.
3.2. Feed distance
The effect of different feed distances on the surface topography and efficiency are shown in Fig. 3, in which the scanning speed of 5 mm/s, the pulse width of 10 µs, and the number of scan 2 are employed. The Roughness, MRR, and the variance S all rise with the feed distance increases, as shown in Fig. 3(a) and (b). Moreover, the MRR increases almost linearly with the increase of the feed distance. The maximum MRR is achieved at the feed distance of 5 µm, while the ablation depth is the minimum which shown in Fig. 3(c). This phenomenon indicates that a large feed distance can improve the MRR while decrease the ablation depth, which means the ablation is insufficient for large feed distance. In order to balance the ablation depth and MRR, it is recommended to increase the pulse energy when using a large feed distance.
3.3 Number of scan
Figure 4 shows the effect of number of scan on the topography of fiber and processing efficiency. The fiber was processed with the pulse width of 10 µs, scanning speed of 5 mm/s, feed distance of 2 µm, and the number of scan ranges from 1 to 5. As shown in Fig. 4(a), both the MRR and the roughness show a downward trend as the number of scan increases. It can also be found that as the number of scan increases, the efficiency decreases, while the ablation depth increases (Fig. 4(c)). Moreover, when the number of scan exceeds 3, the decline in efficiency and roughness will slow down. The variance S shows a decline and then keeps a nearly constant trend, which shown in Fig. 4(b). Due to the dislocation of the spot during the reciprocating scanning process, the protruding part is ablated, while the recessed part will not become deeper. Therefore, the distance Δh between the peak and the valley decreases with the increase of the number of scan, as shown in Fig. 4(d). It also can be concluded from Fig. 4(c) that the ablation depth of flat zone increases with the number of scan increases, while the increasing trend gradually slow down. This means that when the number of scan increases to a certain value, it is not appropriate to remove more material by increasing the number of scan. In addition, when the number of scan exceeds 2, the effect of increasing the number of scan to improve the surface quality is limited. Therefore, the recommended number of scan is 2
3.4 Scanning speed
The effect of scanning speed on topography is demonstrated in Fig. 5. The fiber processed with the pulse width of 10 µs, feed distance of 2 µm, the number of scan 2, and the scanning speed range from 0.5 mm/s to 5 mm/s. It can be concluded from Fig. 5(a) that the MRR and roughness increase with the increase of scanning speed. From Fig. 5(b), it can be seen that S presents a rising-flat-rising trend as the scanning speed increases. This can be explained by the laser spot overlap ratio that is inverse proportional to the scanning speed, as shown in Fig. 5(e). The peak and the valley gap Δz of low scanning speed is smaller than that of high scanning speed, which means low scanning speed can flat the processing surface. It also can be seen from Fig. 5(c) that the fluctuation will gradually disappear with the scanning speed decreasing. However, this does not mean that the scanning speed can be as small as possible, because some molten debris appeared on the surface. Comparing scanning speed of 1 mm/s and 0.5 mm/s, there will be more and more debris, and the size of the debris will become larger and larger, which shown in Fig. 5(d).
4. Discussion and optimization
Based on the above experimental research, we found that fluctuations and molten debris deposition will appear at some specific processing parameters. It is necessary to study the mechanism of these phenomena to optimize the experimental parameters.
4.1 Molten debris deposition
It can be seen from the 3D topography of the surface that the molten debris deposited at low scanning speed and long pulse width. The deposition of debris after laser irradiation on the surface can be described according to the shock wave theory. The plasma stream is generated after laser irradiation on the target, and accelerates outward from the target surface, thereby compressing the surrounding air molecules to form a shock wave. As the shock wave spreads, the pressure in the plume drops, and finally reaches equilibrium with the surrounding air pressure and stops expanding. When the shock wave weakens and the internal gas is cooled by heat conduction and radiation, the ambient gas begins to flow back to the internal area. This inwardly flowing airflow will drag and redeposit debris particles on the surface of the substrate. Moreover, the particle ejection process in the laser ablation is much longer than the pulse duration (on the order of 1-10 us) . The early ejected (smaller in size) particles have a speed of 2-3 km/s , while the particles ejected at the end of the ejected process (which have larger size) having a speed of 10 m/s . With the increase of the laser fluence, the size of the ejected particles also becomes larger . Larger particles may change their trajectory to fall on the surface under the drag force caused by inwardly flowing airflow, and the cooling rate of large particles is slower than small particles . Therefore, when the pulse width (at constant peak power in our case) is increased, more debris will deposit as shown in Fig. 2.
In the case of low scanning speed, debris deposition also appeared on the surface. This is related to heat accumulation . Figure 6(a) and (b) show the maximum surface temperature versus scanning speed and pulse width, respectively. It can be seen from Fig. 6(a) that when the scanning speed is lower than 1 mm/s, the maximum residual temperature of the surface exceeds 680 ℃, which is enough to soften the glass particles and then adhere to the surface. As the scanning speed continues to decrease, the residual temperature gets higher and higher, and more and more debris is melted. Therefore, the debris deposition phenomenon will become more and more serious as the scanning speed decreases as shown in Fig. 5. It also can be concluded that the scanning speed has a more serious impact on the heat accumulation than the pulse width.
4.2 Features of the residual cladding thickness curve
It is worth noting that an oscillation feature in transition area is observed. The oscillation will weaken as the scanning speed decreases, as shown in Fig. 5 (c). The oscillation is caused by fiber vibration. The change of pulling force of the fiber will result in optical fiber vibration. The pulling force is provided by a 100 g movable weight. It can be seen in Fig. 7 that the acceleration changes during the start, turn and stop phases of the movement. When the fiber acceleration changes, there is a delay in the acceleration changes of the weight. Therefore, the pulling force of the fiber changes. The smaller the scanning speed, the smaller the change in acceleration, which results in a small change in the pulling force of the fiber, thus the fiber moves more smoothly.
4.3 Process parameter optimization
Experiments have studied the effects of various experimental parameters on processing efficiency and surface quality. The experimental results show that high efficiency requires large scanning speed, large feed distance and long pulse width, while high surface quality requires low scanning speed, small feed distance and short pulse width. This contradictory requirement means that it is impossible to achieve high efficiency and high surface quality at the same time with one set of parameters. In order to achieve high efficiency and high surface quality, the machining process can be divided into two steps. The first step is to remove most of the cladding material to shorten the processing time. The second part is to improve the quality of the processed surface. Based on the above principles, the first stage gives priority to parameters with high processing efficiency. It can be concluded that the maximum processing efficiency is obtained when the number of scan is 2, the pulse width is 10 µs, the scanning speed is 5 mm/s and the feed distance is 5 µm, as shown in Fig. 3. However, the ablation depth at this time is relatively small, which may be caused by insufficient energy. Therefore, a larger ablation depth can be obtained by increasing the pulse energy, and the processing efficiency will be improved too, such as increasing the pulse width to 35 µs. The second stage is mainly to improve the quality. The experimental results show that for certain parameters such as large pulse width or small scanning speed, the phenomenon of molten debris deposition will occur, which will cause the surface quality to deteriorate. In order to avoid the deposition of molten debris, the feed motion is canceled to reduce the generation of debris, and low fluence pulses are used to avoid heat accumulation. Therefore, the recommended parameters are number of scan 2, the scanning speed of 1.5 mm/s, and the pulse width of 7 us. Based on the above processing parameters, the final manufactured D-shaped fiber is shown in Fig. 8. The RCT of fiber in a flat area maintains within a certain range (134.3 μm -135.7 μm). The roughness of the surface is 46 nm and the variance of RCT is 0.088. The distance from the core boundary to the flat surface is 2.5 μm.
In the present work, an experimental investigation of D-shaped fiber fabricated by a pulse CO2 laser is conducted. The mechanism of defects generation such as oscillation and molten debris deposited on the surface is studied. The oscillation is related to the scanning speed. The lower the scanning speed, the smaller the oscillation amplitude. The mechanism of molten debris deposition is different in the case of long pulse width and low scanning speed. In the case of long pulse width, the number of ejected particles increases with the increase of the pulse width, and the size of particle also becomes larger. The large particles fall back to the surface under the inwardly flowing airflow. With a longer cooling time of the large particles, some particles that have not completely solidified will be adsorbed on the surface. In the case of low scanning speed, the accumulated heat of the surface is too high, which soften the particles fall on the surface. The influence of different processing parameters on processing efficiency and processing quality is also studied. High processing efficiency and high surface quality have contradictory requirements for processing parameters. In order to balance high processing efficiency and high surface quality, the processing process is divided into two stages. The first stage adopts large processing efficiency, such as scanning speed of 5 mm/s, feed distance of 5 μm, the pulse width of 35 μs and the number of scan is 2. The purpose of the second stage is to improve the surface quality. The parameters used in this stage are the scanning speed of 1.5 mm/s, the pulse width of 7 μs and the number of scan 2.
Science and Technology Planning Project of Guangdong Province (2018B090944001); National Natural Science Foundation of China (81927805); Fundamental Research Funds for the Central Universities of HUST (2019kfyXKJC062); DongGuan Innovative Research Team Program (201536000100031); Guangdong Major project of Basic and Applied Basic Research (2019B030302003); Project funded by China Postdoctoral Science Foundation (2018M632837).
The study was supported by the Science and Technology Planning Project of Guangdong Province (Grant no. 2018B090944001); the National Natural Science Foundation of China (NSFC) (Grant no. 81927805); the Fundamental Research Funds for the Central Universities of HUST (Grant no. 2019kfyXKJC062); the DongGuan Innovative Research Team Program (Grant no. 201536000100031); the Guangdong Major project of Basic and Applied Basic Research (Grant no. 2019B030302003); the Project funded by China Postdoctoral Science Foundation (Grant no. 2018M632837). The authors also thank the technical support from Experiment Center for Advanced Manufacturing and Technology in School of Mechanical Science & Engineering of HUST; and the Analytical and Testing Center of Huazhong University of Science and Technology.
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
The data that support the finding of this study are available from the corresponding authors upon reasonable request.
1. H. Chen, S. Li, M. Ma, Z. Fan, and Y. Wu, “Ultrabroad bandwidth polarization filter based on D-shaped photonic crystal fibers with gold film,” Plasmonics 10(5), 1239–1242 (2015). [CrossRef]
2. N. K. Chen, S. Chi, L. Zhang, L. Hu, K. P. Chuang, Y. Lai, S. M. Tseng, and J. T. Shy, “CW-pumped evanescent amplification at 1.55 μm wavelength using highly Er3+-doped glass over side-polished fiber,” in CLEO, (2005), pp. 1352–1354.
3. Y. Luo, Q. Wei, Y. Ma, H. Lu, J. Yu, J. Tang, J. Yu, J. Fang, J. Zhang, and Z. Chen, “Side-polished-fiber based optical coupler assisted with a fused nano silica film,” Appl. Opt. 54(7), 1598–1605 (2015). [CrossRef]
4. H. Chen, S. Li, G. An, J. Li, Z. Fan, and Y. Han, “Polarization splitter based on d-shaped dual-core photonic crystal fibers with gold film,” Plasmonics 10(1), 57–61 (2015). [CrossRef]
5. Y. Ying, G. Y. Si, F. J. Luan, K. Xu, Y. W. Qi, and H. N. Li, “Recent research progress of optical fiber sensors based on D-shaped structure,” Opt. Laser Technol. 90, 149–157 (2017). [CrossRef]
6. P. J. Severin, “Multipurpose single-mode fibre-optic coupling substrate made with silicon etch and polish technique,” Electron. Lett. 25(15), 969–970 (1989). [CrossRef]
7. S. M. Tseng and C. L. Chen, “Side-polished fibers,” Appl. Opt. 31(18), 3438–3447 (1992). [CrossRef]
8. C. D. Hussey and J. D. Minelly, “Optical fibre polishing with a motor-driven polishing wheel,” Electron. Lett. 24(13), 805–807 (1988). [CrossRef]
9. Y. Zhang, L. Wang, and Z. Liu, “The polishing detection method of side-polished fiber,” in Proc SPIE, (2011), p. 820211.
10. S. T. Tso and A. Pask, “Reaction of glasses with hydrofluoric acid solution,” J. Am. Ceram. Soc. 65(7), 360–362 (1982). [CrossRef]
11. D. T. Liang and W. Readey, “Dissolution kinetics of crystalline and amorphous silica in hydrofluoric-hydrochloric acid mixtures,” J. Am. Ceram. Soc. 70(8), 570–577 (1987). [CrossRef]
12. D. J. Monk, D. S. Soane, and R. T. Howe, “A review of the chemical reaction mechanism and kinetics for hydrofluoric acid etching of silicon dioxide for surface micromachining applications,” Thin Solid Films 232(1), 1–12 (1993). [CrossRef]
13. V. R. Machavaram, R. A. Badcock, and G. F. Fernando, “Fabrication of intrinsic fibre Fabry–Perot sensors in silica fibres using hydrofluoric acid etching,” Sens. Actuators, A 138(1), 248–260 (2007). [CrossRef]
14. L. Yin, M. J. Yan, Z. G. Han, H. L. Wang, H. Shen, and R. H. Zhu, “High power cladding light stripper using segmented corrosion method: theoretical and experimental studies,” Opt. Express 25(8), 8760–8776 (2017). [CrossRef]
15. T. He, C. Y. Wei, Z. G. Jiang, Y. N. Zhao, and J. D. Shao, “Super-smooth surface demonstration and the physical mechanism of CO2 laser polishing of fused silica,” Opt. Lett. 43(23), 5777–5780 (2018). [CrossRef]
16. K. Boyd, S. Rees, N. Simakov, J. M. O. Daniel, R. Swain, E. Mies, A. Hemming, W. A. Clarkson, and J. Haub, “High precision 9.6 microm CO2 laser end-face processing of optical fibres,” Opt. Express 23(11), 15065–15071 (2015). [CrossRef]
17. J. R. Grenier, L. A. Fernandes, and P. R. Herman, “Femtosecond laser inscription of asymmetric directional couplers for in-fiber optical taps and fiber cladding photonics,” Opt. Express 23(13), 16760–16771 (2015). [CrossRef]
18. R. Kitamura, L. Pilon, and M. Jonasz, “Optical constants of silica glass from extreme ultraviolet to far infrared at near room temperature,” Appl. Opt. 46(33), 8118–8133 (2007). [CrossRef]
19. H. L. Schick, “A thermodynamic analysis of the high-temperature vaporization properties of silica,” Chem. Rev. 60(4), 331–362 (1960). [CrossRef]
20. G. A. J. Markillie, H. J. Baker, F. J. Villarreal, and D. R. Hall, “Effect of vaporization and melt ejection on laser machining of silica glass micro-optical components,” Appl. Opt. 41(27), 5660–5667 (2002). [CrossRef]
21. R. N. Raman, R. A. Negres, and S. G. Demos, “Kinetics of ejected particles during laser-induced breakdown in fused silica,” (Lawrence Livermore National Lab (LLNL), 2010).
22. S. G. Demos, R. N. Raman, and R. A. Negres, “Time-resolved imaging of processes associated with exit-surface damage growth in fused silica following exposure to nanosecond laser pulses,” Opt. Express 21(4), 4875–4888 (2013). [CrossRef]
23. N. G. Semaltianos, W. Perrie, V. Vishnyakov, R. Murray, C. J. Williams, S. P. Edwardson, G. Dearden, P. French, M. Sharp, S. Logothetidis, and K. G. Watkins, “Nanoparticle formation by the debris produced by femtosecond laser ablation of silicon in ambient air,” Mater. Lett. 62(14), 2165–2170 (2008). [CrossRef]
24. R. N. Raman, S. Elhadj, R. A. Negres, M. J. Matthews, M. D. Feit, and S. G. Demos, “Characterization of ejected fused silica particles following surface breakdown with nanosecond pulses,” Opt. Express 20(25), 27708–27724 (2012). [CrossRef]
25. H. Wang, K. Zhao, H. Shen, and Z. Q. Yao, “Experimental study on direct fabrication of micro channel on fused silica by picosecond laser,” J. Manuf. Process. 55, 87–95 (2020). [CrossRef]