Open Access
Add to BibSonomy
Abstract
We detailed the design and fabrication of a beam quality preserving and low insertion loss asymmetric (6 + 1)x1 pump and signal fiber combiner. Transmission efficiency of the beam combiner with different structure parameters such as taper ratios, taper lengths, etc are analyzed using the beam propagation method numerically. We experimentally demonstrated and investigated three design parameters and fabrication methods, including chemical wet etching, pre-tapering of fibers, and thermally expanding the core of bundle. Negligible beam quality factor Mx,y2 degradation of 0.05 / 0.02 and low insertion loss of 0.02 dB for pump and 0.16 dB for signal ports respectively were achieved.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Fiber beam combiner is one of the critical components for monolithic high-power fiber laser and amplifier systems. Many research groups worldwide are developing fiber combiners with different structures and architectures for high-power fiber laser systems [1], kW fiber amplifiers [2], super continuum generation [3] and etc. Some worked on feeding more pump fibers into the combiners [4–7], while others focused on high transmission efficiency [8,9]. D. Stachowiak gave a comprehensive summary on the recent achievements of high-power passive fiber components, including pump and signal combiners [10]. He discussed the basic operation principles and various fabrication techniques of beam combiners in detail. Beam quality is an important performance indicator of fiber lasers and amplifiers. It describes how well the laser power can be delivered to a certain area. Therefore, signal beam quality preservation of the pump and signal (N+1)x1 fiber beam combiners should be emphasized for good performance and power scaling capabilities of the systems.
J. Zheng, et. al [2] demonstrated a 2.67 kW fiber amplifier using their (6 + 1)x1 backward fiber beam combiner; achieved good beam quality of 1.41 and average pump coupling efficiency of 94.83%. K. Liu, et. al [11] presented a low signal beam quality M2 degradation of ∼ 0.15 in their symmetric (6 + 1)x1 pump and signal fiber beam combiner, however at a reduced signal efficiency of ∼ 87.5% with a built-in mode field adapter. The team later demonstrated two asymmetric (6 + 1)x1 fiber combiners with six pump fibers of 220/242 µm, a signal fiber of 30/250 µm and an output fiber of 20/400 µm [12]. They employed (i) coreless intermediate fibers and (ii) ultra-long pretapered pump fibers in conjunction with a chemically etched signal fiber in their fiber bundles; and achieved signal efficiency of ∼ 95%, pump efficiency of ∼ 98% and signal beam quality M2 degradation of 0.32–0.64. Y. Gu and team [13] also showed an asymmetric (6 + 1)x1 beam combiner with transmissions of ∼ 98% and signal beam quality degradation of 0.78–1.24 using a signal transition fiber in their combiner. Besides the tapered fiber bundle end pumped architecture, C. Lei and team [14] worked on the tapered fused side-pumped fiber combiners. They showed that the beam quality degradation and signal insertion loss are dependent on the temperature and duration of the heating process.
Symmetricity of the beam combiner describes the matching of pump and signal fibers. We define the symmetricity factor, s to be the ratio of cladding diameters of signal fiber, ds, to pump fiber, dp, i.e $s = {d_s}/{d_p}\; $ for a standard (6 + 1)x1 pump and signal fiber combiner. The combiner is considered symmetric if s = 1. Symmetric fiber combiner, as the name suggests, gives better symmetricity, and hence centering of the signal fiber in the bundle. However, it limits the choice of fibers. Asymmetric fiber bundle, on the other hand, allows for a wider selection of fibers but potential centering and misalignment issues.
In this paper, we designed, fabricated, and investigated the asymmetric (6 + 1)x1 pump and signal fiber beam combiner with the following fiber configurations:
- (i) Six pump fibers: multimode fiber with core / cladding diameters = 135 / 155 µm, NA = 0.22 (MMF-GDF-135/155-M)
- (ii) One signal fiber: double clad fiber with core / cladding diameters = 25 / 250 µm, core / cladding NA = 0.065 / 0.46 (LMA-GDF-25/250-M)
- (iii) Output fiber: double clad fiber with core / cladding diameters = 25 / 250 µm, core / cladding NA = 0.065 / 0.46 (LMA-GDF-25/250-M)
The symmetricity factor s of the proposed structure is 1.61, which is much larger than the one reported in [12,13], i.e. ∼ 1.04. This increases the matching demand of fiber bundle. We analyzed the pump and signal transmission behaviors numerically and selected three sets of fabrication parameters accordingly. Both pump and signal fibers were pre-processed either by chemical etching or pre-tapering to fulfill the fabrication criteria. The designs are described and investigated in the subsequent sections. Using the pre-tapered pump fibers, chemically etched signal fiber, and thermally expanded core of the fiber by CO2 laser; we obtained the pump and signal transmissions of ∼ 99.5% and ∼ 96.4% respectively. The signal beam quality M2x,y degradation was measured to be 0.05 / 0.02. To the best of our knowledge, this is the smallest beam quality degradation reported to-date.
We believe that the main contributing factor to achieve this low loss and low beam quality M2 degradation combiner is the use of clean and stable CO2 laser as the heat source for the fabrication. The laser leaves no deposit or contamination on the fiber, hence provides ultra-low loss and good matching conditions for the combiners. Additionally, smooth lateral etch profiles by our proposed etchant also aids the success of the combiners.
2. Theoretical analysis
We investigated the transmission behaviors of the (6 + 1)x1 pump and signal fiber beam combiner using beam propagation method (BPM) under MATLAB environment, i.e. BPM-MATLAB, an open-source optical simulation tool in MATLAB. It is a numerical simulation tool in which the Douglas-Gunn Alternating Direction Implicit (DG-ADI) method is used to efficiently model the electric field propagation using a Finite-Difference Beam Propagation Method (FD-BPM) in a wide variety of optical fiber geometries with arbitrary refractive index profiles [15].
The following combiner structure was created: six input pump fibers (MMF-GDF-135/155-M) and a signal fiber (LMA-GDF-25/250-M) were tapered, bundled, and spliced to an output fiber (LMA-GDF-25/250-M). The influence of taper ratios and taper lengths to the transmission efficiency were investigated. We define the taper ratio as the ratio of initial to waist dimensions, and internal separation as the distance between the two opposite pump fibers - it represents the cladding diameter of signal fiber in the structure. The assumptions and parameters used in the simulation are tabulated in Table 1. A typical fiber combiner structure and its beam propagation behavior are shown in Fig. 1.
Fig. 1. Typical (a) structure and (b) beam propagation behavior of (6 + 1)x1 pump and signal fiber beam combiner based on linear taper profile.
Table 1. Assumptions and parameters used in the analysis
Linear taper profile is selected for simulation due to its simplicity and constant taper slope throughout the taper length. The fabricated taper may differ slightly from the desired case. The transmission loss is minimal for taper profiles with taper angle smaller than the small angle limit everywhere along the taper, known as adiabatic criterion of a fiber taper [16,17]. Therefore, the taper profile plays a less determining role in the transmission analysis if the adiabatic criterion is satisfied. The analysis of the various taper profiles is beyond the scope of this work.
The pump transmissions for the four different internal separations are depicted in Fig. 2. The legends denote the taper lengths used in the calculation. Signal fiber is processed to the dimensions indicated by the internal separations. Internal separation of 250 µm represents the original pump and signal fibers. This is the most primitive structure of the bundle; it shows the lowest pump transmission among the four. For internal separations of 125 and 85 µm, the pump fibers are either pre-tapered or etched down to that of the dimensions for symmetricity matching. Based on the analysis, the pump transmission goes beyond 98% as long as (i) the taper length is longer than 8 mm and (ii) taper ratio is slightly larger than the optimum taper ratio defines as TR$= ({{d_i} + 2{d_p}} )/{d_o}$, where di, dp and do are the internal separation, pump fiber diameter and output fiber diameter respectively.
Fig. 2. Pump transmission for internal separations of (a) 250 µm, (b) 155 µm, (c) 125 µm, and (d) 85 µm. The legends show the taper lengths of the bundle
The signal transmission efficiency of a truly single or fundamental mode beam is shown in Fig. 3(a). It decreases with the taper ratio, however less affected by the taper length. To achieve signal transmission of > 95%, the taper ratio has to be < 2.4. We focused on the taper length of 10 mm for the subsequent signal analysis to match the pump transmission efficiency.
Fig. 3. Signal transmission of (a) truly single mode signal, legend shows the taper length, (b) different signal mode compositions with taper length = 10 mm. Legend shows the number of modes involved in the analysis
We included some high order modes into the signal transmission analysis. The distributions of the modes are by weightage, however with random phase. The result is presented in Fig. 3(b). The signal transmission deteriorates drastically with the number of modes and taper ratios. One conclusion that we can draw is that the input signal condition also contributes to the transmission efficiency of the combiner, not merely by the physical structure of the component.
Based on the analysis, two essential criteria for low loss beam combiner operations are: (i) taper length longer than 8 mm, and (ii) taper ratio lesser than 2.4. Thus, three sets of parameters were selected for fabrication, that is taper ratios around 1.86, 1.50 and 1.02 with taper length of 15 mm. The fabrication methods will be further described in the subsequent section.
3. Fabrication methods
The main process used in the fiber beam combiner fabrication is basically the fiber thinning process, either by tapering or etching. Tapering shrinks both the core and cladding dimensions of the optical fiber. It is done by heating the fiber to its softening point and stretching it. The softening point for silica fibers is around 1600°C. The heat source used in the process can range from a hydrogen flame, resistively heated metal or graphite filament, electric arc to CO2 laser. Electric arc discharge heating is the most used method, where a voltage is applied across two or three electrodes separated by an air gap, the resulting current flow will heat up the fiber. Conversely, laser heating utilizes the strong silica absorption at 10.6 µm of the CO2 laser. We used CO2 laser glass processing station for this work. The system employs a 30 W CO2 laser that provides a clean and stable operation. However, the laser may require gas recharge every 3 to 5 years of operation.
Etching, on the other hand reduces only the outer cladding of the fiber and leaves the core untouched. It removes the material from outer surface of optical fiber by chemical reaction [18]. Acetic acid is often used for etching the plastic fibers while hydrofluoric acid (HF) for silica fibers. Here, we propose to use glass etching solution for the work. HF acid, being extremely toxic and corrosive, can cause serious safety issues and even death; requires special trainings on the use of it. In contrast, the glass etching solution is safer and easier in handling, though it is still corrosive in nature. Hence, basic safety requirement is still required.
Here we used glass etching solution Etchall dip ‘n etch with ∼ 27% Ammonium Bifluoride for this work. Ammonium bifluoride or ammonium hydrogen fluoride (NH4HF2) is a colorless inorganic compound produced from ammonia and hydrogen fluoride. It attacks silica component of glass with action stated in Eq. (1).
We investigated the etch rate of solution with concentrations of 27%, 18%, 9% and 4.5%, and the results are charted in Fig. 4. The etch rates vary from 0.42 µm/min (for 27% etchant) to 0.05 µm/min (for 4.5% etchant). It takes about 4 hours (for 27% etchant) to more than 30 hours (for 4.5% etchant) to etch a 250 µm fiber to 150 µm. The result is consistent and repeatable. Although the process is time consuming, it provides better controllability and resolution compared to 49% HF acid.
We also investigated the surface roughness of the etched fibers by the different concentrations, as shown in Fig. 5. Crystal-like or pebbly structures are formed on the etched surfaces of the high concentration etchant, while smooth finishing is observed for the low concentration ones. We also found that the fibers etched with concentrated etchant appeared to be more brittle. One thing worth mentioning is that no additional fire polishing is needed for fibers etched using diluted etchant, however day(s) of etch time. As a result, we chose 4.5% etchant for this work for precision and good surface roughness.
In addition to the surface roughness, we also observed a lateral slope profile at the boundary of etchant and fiber as shown in Fig. 6, instead of a step profile. The slope etch profile is mainly due to the adhesive force of the solution to the cladding of fiber.
In standard fiber beam combiner fabrication, the capillary tube will first be tapered down to accommodate the bundle dimension. Six pump fibers and an input signal fiber with stripped coating are inserted to the tapered tube to create the fiber bundle. It is then cleaned using an ultrasonic acetone bath and heat-dried or vacuum-dried to evaporate the residual solvent and moisture. The drying step is crucial to avoid ‘carbonization’ during the main taper process. The bundle is later tapered down to match the dimension of the output fiber. Finally, it is cleaved and spliced to the desired output fiber. However, for this beam quality preserving and low loss asymmetric pump and signal combiner, pre-processing of the fibers as mentioned earlier is needed.
Three design parameters are selected as below. Low index fluorine doped fused silica tubes with inner / outer diameters of 800 / 1100 µm, and NA of 0.22 were used throughout this work.
- • Design #1: Signal fiber is etched to size of pump fibers, i.e. 155 µm. Taper ratio ∼ 1.86–2.56*
- • Design #2: Signal fiber is etched to 125 µm and pump fibers are pre-tapered to 125 µm. Taper ratio ∼ 1.50–2.06*
- • Design #3: Signal and pump fibers are etched to 85 µm. Taper ratio ∼ 1.02–1.40*
4. Results
Signal fibers were chemically etched to the required dimensions for Design #1 to #3 using the 4.5% etchant. Unprocessed pump fibers were used in Design #1, tapered pump fibers in Design #2 and etched pump fibers in Design #3.
The end faces of the fiber bundles for three different designs are shown in Fig. 7. The cladding of the unprocessed and tapered pump fibers can be easily identified in Design #1 and Design #2. While the etched pump fibers have been homogenized and resulted in paddle-like inner cladding in Design #3.
The refractive index profiles of the fiber beam combiner were measured using IFA100 index profiler. The fringe pattern near the interface of tapered fiber bundle and output fiber; as well as the refractive index profile of the fiber bundle are shown in Fig. 8. The measurement shows that the NA of the low index tube used is around 0.22, which matches well with the cladding of pump fibers.
The setup of measurement is depicted in Fig. 9. A 30-Watt 915 nm fiber coupled laser diode with NA of 0.22 was used for pump port evaluations. Meanwhile, a homemade 10-Watt Ytterbium doped fiber laser (YDFL) operating at 1080 nm was used for signal port measurements. Although we did not use the cladding mode stripper in our laser, most of the cladding light was stripped at the 20/400–25/250 connection shown in Fig. 9(b). The LMA25/250 was tapered down to 200 µm for better mode field diameter matching with the LMA 20/400 fiber. Based on the beam measurement of the YDFL, it has more than one mode exist in our system, however with unknown compositions. The signal beam quality M2 factor was measured using Thorlabs M2MS - M2 measurement system with scanning slit beam profiler. Please note that power and beam quality referencing (Pref and M2ref) were taken after the splice point A. This is to exclude the potential power loss and beam degradation due to mismatch at the splice point.
Fig. 9. Experimental setup for (a) pump port and (b) signal port evaluations. Red dots refer to the splice points, DUT: device (fiber combiner) under test, M2: M2 measurement system.
We have successfully fabricated and investigated 17, 23 and 5 pieces of Design #1, Design #2, and Design #3 fiber combiners respectively – which brings a total of 45 fiber combiners. The performance of the combiners based on the best effort is tabulated in Table 2. The insertion loss distributions for Design #1 and Design #2 are shown in the box and whiskers plot in Fig. 10, with inset shows the zoom-in results. The box and whiskers plot shows the loss distributions through their quartiles.
Fig. 10. Insertion loss distribution for Design #1 and Design #2. Inset shows the zoom-in loss from 0.1 to 0.6 dB.
Table 2. Insertion loss and signal beam quality degradation of Design #1 to Design #3 beam combinersa
Out of the three designs proposed and implemented, Design #3 produced the most fragile structure and hardest to handle, although it gives the best results theoretically. Further, some ports were burnt during the insertion loss measurement. The survival rate for Design #3 is very low. We believed that the fragility is primarily due to the high etch volume of fibers, i.e. from 250 µm and 155 µm to 85 µm; i.e. about 90% and 70% of the volume were etched away for the signal and pump fibers respectively.
Design #1 is the strongest physically, easiest to handle and it gives relatively low loss performance, i.e. pump loss of < 0.1 dB and signal loss of < 0.3 dB. Nonetheless, the signal beam quality degradation of ∼ 0.3 due to higher taper ratio.
Design #2 stands in as a better option among the three. It is mechanically stronger than Design #3, and theoretically better performance than Design #1. However, this design requires more optimizations. Pump fibers were pre-tapered to 125 µm and signal fiber was etched to the same dimension prior to the bundling process.
We achieved the best results using Design #2. Pump and signal transmissions of 99.5% and 96.4% respectively. The results agree well with the theoretical analysis for pump transmission with internal separation of 125 µm and signal transmission between single and two modes beam, both at taper ratio of ∼ 1.9. The design also well preserves the signal beam quality due to smaller taper ratio compared to Design #1. Negligible signal beam quality M2 degradation of around 0.05 / 0.02 was obtained. The theoretical signal loss for single and two modes beam is depicted in Fig. 11. The experimental signal insertion loss for Design #1 and Design #2 are marked as red crosses in the figure, are within the yellow and blue regions respectively. The spans of regions are due to the thickness of capillary tubes and some uncertainties in the fabrication. The results show a closer match to the analysis near the single mode beam.
The signal beam quality M2 factor was measured and is presented in Fig. 12. Thorlabs M2MS-BP209 uses the beam propagation and principle of scanning slits for the M2 measurement according to the ISO 11146 standard. Two orthogonal slit pairs are mounted on the perimeter of a rotating drum, which can be adjusted to adapt to the major and minor axes of an elliptical beam. The drum should be adjusted to a point where the profile width reaches minimum in one axis and maximum in the other for accurate results, otherwise erroneous measurement can be expected. M2 degradation between 0.02 to 0.34 was observed for Design #1 and Design #2. The degradation is mainly due to the mode field mismatch between the bundle and output fiber, potential misalignment and imperfection during splice, or / and imperfect alignment of the measurement setup.
Fig. 12. M2 measurement of (a) signal, with the inset shows the beam profile; (b) reference beam; for Design #2 beam combiner
The highest achieved output power was about 30 W due to limited pump power available in our lab. At this pump power, the beam combiner dissipates about 0.14 W of heat (based on 0.02 dB pump insertion loss), minor temperature-rise of about 0.8°C was observed under no thermal management, i.e. 0.03°C / W of temperature rise, as shown in Fig. 13. Based this temperature rise, the combiner is capable of higher power operations.
Fig. 13. Thermal images of the beam combiners. (a) background temperature; (b) beam combiner temperature when ∼ 30 W of laser power is injected.
5. Discussion
The main trick that we employed in making this beam preserving and low insertion loss fiber combiner is the multiple refiring technique at the splice point between the fiber bundle and output fiber. The multiple refiring of the CO2 laser beam not only allows for low splice loss, but also diffuses the core dopants and hence expands the core of the fiber bundle, i.e. thermal core expansion. This provides a better mode field diameters matching between the bundle and output fiber and improves the performance of the combiner. However excessive re-firing can cause significant tapering and bulging at the joint, which will in turn increase the splice loss, make the taper less durable and more susceptible to premature failure.
Tighter dimension control, be it etched fiber, tapered fiber, tapered tube, or taper profile is another critical requirement in making this combiner a success. Otherwise, misalignment or non-circular fiber bundle as shown in Fig. 14 will cause undesirable operations.
6. Conclusions
Three sets of design and fabrication parameters have been identified and implemented for the asymmetric (6 + 1)x1 pump and signal fiber beam combiner. The characteristics of the designs are concluded in Table 3. Design #2: pre-tapered pump and etched signal fibers to 125 µm, produced the best results of the three. We achieved low insertion loss of 0.02 dB for pump and 0.16 dB for signal ports respectively; and minor signal beam quality Mx,y2 degradation of 0.05 / 0.02. Temperature-rise of 0.03°C / W was observed under no thermal management. Based on this thermal behavior, the combiner is capable of higher power operations.
Table 3. Design summary
Funding
Nanyang Technological University (DSOCL19195).
Acknowledgments
We would like to thank the laboratory managers and technical support officers of COEB, COFT, OPTIMUS and TPI at NTU for the support given and the use of their equipment.
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. W. Yao, C. Shen, Z. Shao, J. Wang, F. Wang, Y. Zhao, and D. Shen, “790 W incoherent beam combination of a Tm-doped fiber laser at 1941nm using a 3 ( 1 signal combiner,” Appl. Opt. 57(20), 5574–5577 (2018). [CrossRef]
2. W. Li, C. Hou, P. Wu, S. She, W. Gao, P. Ju, Q. Gao, G. Li, C. Chang, Z. Li, B. Zhao, W. Zhao, and J. Zheng, “High pumping-power fiber combiner for double-cladding fiber lasers and amplifiers,” Opt. Eng. 57(03), 1 (2018). [CrossRef]
3. H. Zhou, A. Jin, Z. Chen, B. Zhang, X. Zhou, S. Chen, J. Hou, and J. Chen, “Combined supercontinuum source with >200 W power using a 3 ( 1 broadband fiber power combiner,” Opt. Lett. 40(16), 3810–3813 (2015). [CrossRef]
4. M. Faucher, B. Sevigny, R. Perreault, A. Wetter, and N. Holehouse, “All-Fiber 32 × 1 Pump Combiner with High Isolation for High Power Fiber Laser,” in CLEO/QELS, (2008).
5. B. H. Kim, S.-J. Kim, Y. Yoon, and S. Hann, “Fabrication of the reliable (14-18)×1 fiber laser power combiner by the novel double bundling method,” in Optical Components and Materials X, (2013).
6. M. H. Muendel, R. Farrow, K.-H. Liao, D. Woll, J. Luu, C. Zhang, J. J. Morehead, J. Segall, J. Gregg, K. Tai, B. Kharlamov, H. Yu, and L. Myers, “Fused fiber pump and signal combiners for a 4-kW ytterbium fiber laser,” in Fiber Lasers VIII: Technology, Systems, and Applications, (2011).
7. H. Zhanga, C. Leia, Z. Lia, and Z. Chen, “High power tapered fiber bundle 19 × 1 pump combiner,” in Asia Communications and Photonics Conference, (2020).
8. D. Stachowiak, P. Kaczmarek, and K. M. Abramski, “(5 + 1) × 1 pump and signal power combiner with 9/80 µm feed-through signal fiber,” Opt. Laser Technol. 93, 33–40 (2017). [CrossRef]
9. S. Zou, H. Chen, H. Yu, J. Sun, P. Zhao, and X. Lin, “High-efficiency (6 + 1) × 1 pump–signal combiner based on low-deformation and high-precision alignment fabrication,” Appl. Phys. B 123(12), 288 (2017). [CrossRef]
10. D. Stachowiak, “High-Power Passive Fiber Components for All-Fiber Lasers and Amplifiers Application—Design and Fabrication,” Photonics 5(4), 38 (2018). [CrossRef]
11. K. Liu, C. Zhao, Y. Yang, X. Chen, J. Wang, B. He, and J. Zhou, “Low beam quality degradation, high-efficiency pump and signal combiner by built-in mode field adapter,” Appl. Opt. 56(10), 2804–2809 (2017). [CrossRef]
12. Y. Liu, K. Liu, Y. Yang, M. Liu, B. He, and J. Zhou, “High power pump and signal combiner for backward pumping structure with two different fused fiber bundle designs by means of pretapered pump fibers,” Opt. Express 29(9), 13344–13358 (2021). [CrossRef]
13. Y. Gu, C. Lei, H. Yang, H. Xiao, J. Leng, and Z. Chen, “High-beam-quality signal and pump combiner with large-mode-area fiber for high-power fiber laser and amplifier,” Appl. Opt. 58(6), 1336–1340 (2019). [CrossRef]
14. C. Lei, Z. Chen, H. Yang, Y. Gu, and J. Hou, “Beam quality degradation of signal light in a side pumping coupler with a large-mode-area signal fiber,” Opt. Express 27(10), 14041–14053 (2019). [CrossRef]
15. M. Veettikazhy, A. Kragh Hansen, D. Marti, S. Mark Jensen, A. Lykke Borre, E. Ravn Andresen, K. Dholakia, and P. Eskil Andersen, “BPM-Matlab: an open-source optical propagation simulation tool in MATLAB,” Opt. Express 29(8), 11819–11832 (2021). [CrossRef]
16. T. A. Birks and Y. W. Li, “The Shape of Fiber Tapers,” J. Lightwave Technol. 10(4), 432–438 (1992). [CrossRef]
17. J. D. Love and W. M. Henry, “Quantifying loss minimisation in single mode fibre tapers,” Electron. Lett. 22(17), 912–914 (1986). [CrossRef]
18. C. Iliescu and J. Miao, A review on wet etching of glass, Glass Materials Research Progress (Nova Science Publisher, 2008).
