A novel method for assembling endcaps to optical fibers is presented. The method relies on femtosecond laser welding and milling of a glass slide to the polished end of the fiber. The fiber is welded to the glass slide in the cladding region so as to seal the core area without affecting its optical transparency. The same laser is used to mill through the glass slide thereby shaping a microscopic endcap with a diameter slightly larger than that of the fiber. The method was applied to both a standard and a microstructured optical fiber. Preliminary results are also presented on femtosecond laser welding parallel to an interface showing the potential of this approach for optical fiber fusion splicing.
© 2013 Optical Society of America
Optical fiber end-caps are used to protect or to seal optical fiber tips in various situations. In fiber lasers, endcaps are used to increase the damage threshold of the component. Alternatively, endcaps can also serve to reduce the unwanted Fresnel reflection from the fiber end face. Fiber endcaps can also be used to seal the air-holes of microstructured optical fibers (MOF) used in the generation of supercontinuum  and thus prevent their progressive degradation due to OH diffusion inside the glass.
The standard approach to assemble an endcap consists in fusion-splicing a coreless silica rod to the fiber end face [2,3]. The rod can afterwards be tapered, polished, cut or shaped by laser ablation and/or chemical etching to modify its thickness or length. Fiber fusion splicing can be performed by locally heating the parts to be assembled through the use of an electric arc discharge, a narrow omega-shaped resistive (e.g. tungsten) filament or a CO2 laser [3–5]. In all the preceding approaches, the fusion splicing involves significant melting of the fiber tips that inevitably deform to a certain extent the core area, and is thus likely to degrade the quality of the transmitted beam. Although commercial fusion splicers have been optimised for standard fibers, there are several situations involving specialty fibers where low-loss splices cannot be currently achieved due to this intrinsic core deformation problem. For instance, in hollow core photonic crystal fibers (PCF), any appreciable heating of the fiber end is likely to significantly alter the fiber delicate periodic structure and thus the supported mode. As an alternative to fusion splicing, the fiber may be inserted into a capillary, the latter being bonded to an optical flat . Although this approach does not induce melting of the fiber and seals it from contamination, adhesives are necessary to hold the assembly together and therefore has a limited life span and would not support high temperatures.
In this work, we propose the use of femtosecond (fs) laser micromachining techniques to assemble an endcap on the tip of both a standard silica fiber as well as a microstructured optical fiber (MOF). The method we propose first involves a careful polishing of the fiber tip which is subsequently put in optical contact with the surface of a flat and thin glass slide. The fiber is then welded to the glass slide by transmission laser welding [7–10]. The major benefit of using fs laser pulses is that the interaction and thus the heat transfer to the material is confined solely to the focal volume through non-linear excitation processes . Therefore, welding the materials in the cladding area of the fiber has no effect on the core or air-hole regions. To the best of our knowledge, this is the first demonstration of ultrashort laser welding of the tip of optical fibers to a glass substrate. Following the welding process, the same laser is used to mill through the thin glass slide so as to shape an endcap with a diameter slightly larger (15 to 30 µm) than that of the fiber. Ideally, the endcap remains in optical contact with the fiber after the process in such a way that no double Fresnel reflections are generated at the fiber-endcap interface. The method has been demonstrated for both standard and MOF fibers. Preliminary results are also presented on femtosecond laser welding parallel to an interface showing the potential of this approach for optical fiber fusion splicing.
2. Assembling of an optical fiber end cap by transmission laser welding
2.1 Experimental procedure
Femtosecond laser welding of glass can be achieved by focusing a high intensity laser beam at normal incidence across an interface. This so-called transmission laser welding requires that the gap between the joining surfaces is of the order of λ/4 or less [7–9]. This was first achieved on glass slides by applying external forces to the assembly . However, it was recently shown that the most effective way to reduce the gap to less than λ/4 over large areas is to establish optical contact (i.e. direct bonding) between them [8,9]. Therefore, the first step in assembling an endcap to optical fibers is to carefully polish and clean the fiber tip. The flat fiber tip can then be put in optical contact with a glass substrate and welded thereupon by irradiating the cladding area following a circular pattern so as to seal the core area. Finally, the same laser is used to cut the glass around the fiber circumference. These steps will be described in greater detail in this section.
2.1.1 Fiber polishing
Fiber preparation is executed by inserting the uncoated end into a ceramic ferrule and then rubbing the ferrule-fiber assembly on diamond abrasive polishing mats so as to obtain a flat end face. The fiber is fixed inside the ferrule with melted wax. The assembly is then polished sequentially on 5 different mats, starting with a 15 µm grit size and then diminishing the grit size, the finishing step being executed on a 1 µm grit size mat. The ferrule-fiber assembly is then heated to remove the fiber from the ferrule. Finally, the end face is cleaned by dousing the fiber tip in iso-propanol for 1 hour in an ultrasonic bath. Figure 1shows an optical microscope image and a surface profile measurement of the polished tip of a 240 µm diameter fiber.
The roughness of the end face can be estimated to be below 5 nm, which corresponds to the resolution of the interferometer used for the surface profile measurements (Data Pixel 3D scope fiber interferometer). The flatness of fibers polished according to this method is well below the λ/4 requirement for successful welding (λ/4 ≈150 nm height difference).
2.1.2. Achieving optical contact
Optical contact is obtained by aligning the fiber tip parallel to the glass slide surface. If the tip is not well prepared or misaligned with respect to the plate surface, optical contact may form on only a portion of the fiber’s surface area, as shown by the dark spot in Fig. 2(a).Lift-off of optical contact is observed through the colored fringes. When these faults are alleviated, the dark spot and therefore optical contact extends to cover the whole fiber surface, as demonstrated in Fig. 2(b).Weak Van der Waals forces attract the two surfaces together in the optically contacted area so as to form a direct bond . However, due to its small surface area, this bond will break if having to support the weight of either the fiber or the glass plate. Therefore, reinforcing this bond by fs laser welding will not only solidify the direct bonded area but also maintain the state of optical contact between the fiber and the glass slide .
2.1.3 Femtosecond laser welding and sealing
Focussing of high intensity fs laser pulses inside a transparent material was shown to initiate non-linear absorption processes [11,13]. For loose to moderate focusing conditions, the intensity at the focal spot will be mainly governed by the optical filamentation phenomena which is arising from a balance between Kerr self-focusing and plasma defocusing . The local plasma generated by this nonlinear absorption will in turn induce a permanent structural change of the material in the filament volume . When an optical filament is crossing the direct bonded area between two transparent materials, a physical link (i.e. weld) can be produced as a result of the combined structural changes occurring in both materials at the interface. By translating the sample with respect to the beam focus a weld seam can thus be written [7–10]. Filamentation based fs laser welding was also shown to be an efficient mean to reinforce direct bonded transparent components [8,16]. Weld seams are written by multiple scans in a closed shape pattern at the outskirts of a direct bonded region so as to seal the central area, thereby defining an optical transmission window. Accordingly, the optical properties of the encircled window are not altered by the bonding process. The mechanical strength of the assembly is also enhanced compared to unreinforced direct bonded components. Figure 3 illustrates this process when applied to a fiber-glass slide assembly. Femtosecond laser pulses are focused through the glass slide so as to produce optical filaments at the fiber-slide direct bonded interface. The assembly is translated in a spiral motion so as to write continuous curved weld lines in the cladding area, the resulting optical transmission window being positioned at the fiber’s core. With the optical contact maintained inside the sealed area and the material modification localized to the exposed region, the resulting joint does not affect the optical transmission of the fiber mode.
The welding setup is essentially identical to the one previously introduced in  and . We used a REGA 9050 fs laser system from Coherent Laser Group Inc. for all experiments, delivering pulses with a duration of 70 fs with spectral bandwidth centered at 787 nm. The repetition rate was set at 250 kHz yielding a maximum pulse energy of 4 µJ. The laser was focused inside the samples with two different AR coated microscope objective lens from New Focus (5X, 0.1 NA and 10X, 0.16 NA). Scanning the laser inside the samples is achieved with the help of a Newport XPS three axis translation stage system. For welding experiments with fibers, a fiber alignment device was incorporated to the setup to allow precise alignment of the fiber tips with the glass plates.
2.1.4 Femtosecond laser milling of the endcap
Multiple strategies have been proposed to cut transparent materials with ultrashort laser pulses. Undoubtedly, the best approach is to mill from the rear surface to the front surface thereby insuring consistent volumetric absorption and a smoother ablated material plume expansion . Nonetheless, cutting with low energy fs laser pulses (i.e. a few µJ) requires a material removal strategy otherwise the ablated material will redeposit inside the kerf when cutting is to be executed in ambient conditions. Therefore, we used the liquid assisted rear surface drilling concept to mill the endcap [17,18]. In doing so, the ablated material dissolves in water during milling and does not redeposit in the kerf, which allows deep drilling. Following the welding of the fiber, the plate’s rear surface is put in contact with water with the laser focus position at the water-glass interface, as demonstrated in Fig. 4(a). Upon irradiation, the assembly is moved following a trepanning movement with a diameter slightly larger than that of the fiber so as to cut through the glass plate up to the front surface (Fig. 4(b)).
Figure 5 shows a scanning electron microscope (SEM) image of a 100 µm thick fused silica endcap assembled to a microstructured optical fiber. The endcap thickness corresponds to that of the glass slide and its diameter is 30 µm larger than the fiber’s diameter. We note that the fabrication of such a short (i.e. thin) end-cap would have never been possible with the standard fusion splice and cleave approach.
2.2 Experimental results
2.2.1 Femtosecond laser welding of a glass slide to the end face of standard optical fibers
Experiments were first performed on two types of standard multimode fibers from Thorlabs, a UM22 fiber with 200/220 µm core/cladding diameters and a BFL37 with 400/430 µm core/cladding diameters. All fiber samples were cut into 10 cm long segments. For each segment, the jacket was removed at one end and the bare tip was polished according to the method described in section 2.1. The fused silica glass slides used for this experiment were 700 µm thick and cut into 10 x 5 mm substrates.
Proof of concept following the welding method described in section 2.1 was first performed with BFL37 400 µm diameter silica rods. Following the achievement of optical contact between the fiber tip and glass slide, an annular welded region was written by 30 passes of the laser spaced 2 µm apart in a spiral like motion so as to seal off an inner 200 µm diameter optical transmission window. Figure 6 shows an optical microscope image of the tip of the BFL37 fiber after welding where optical contact is maintained between the joining surfaces.
To evaluate the joint’s resistance to tensile stress, several BFL37 fibers were joined in the same manner as the one shown in Fig. 6. Additionally, multiple segments of UM22 fiber were also joined by writing 57 weld lines spaced 2 µm apart with a 10 µm diameter optical transmission window. The tensile strength at breakage of the joints was measured by fixing the glass slide to a table and attaching the unprocessed end of the fiber to a dynamometer (vertical scale). A load of approximately 1 gram/second was applied by slowly pulling on the assembly with a string attached to a translation stage. The tensile strength (in MPa) was then calculated by dividing the measured load at breakage by the area of the annular weld. Each MPa of tensile strength corresponds to applied loads of 5.0 gr (BFL37) and 3.9 gr (UM22). Figure 7 presents results for both types of fibers. For the UM22 fiber, three pulse energies were used.
Most fiber-glass slide assemblies broke in the 5 to 15 MPa range, yielding mean values for each group of measurements ranging between 8.2 and 11.1 MPa. Note that neither the type of fiber nor the weld geometry seemed to significantly affect the tensile strength. For the UM22 fiber, the measurements demonstrate a slight increase in mean tensile stress at breakage with increasing pulse energy although the data at high energy seem to be more scattered. Nishii et al. observed a similar trend when welding fused silica blocks using a similar method . The breakage stress of one assembly overshot the others with a breakage stress of 33.6 MPa. This value was not considered in the calculation of the mean stress for group UM22 600 nJ. Although this measurement was not repeated in our experiment, it demonstrates the potential of using ultrashort laser pulses for welding applications when ideal conditions are brought together. Since our experiment involves multiple steps prior to welding, several factors could have influenced the welding and breakage test conditions. We suspect that our maximum value was obtained in ideal conditions, where the surface and weld seam qualities were superior and the pull test did not induce undesirable shear and torsion stress to the joint.
Figure 8 shows the peeled off joined surfaces of one of the glass slide-UM 22 fiber assemblies used for the tensile strength measurements. The glass slide broke around the welded region thereby leaving a chunk of material on the fiber tip. Similar observations were made when welding fused silica blocks by filamentation . Therefore, we suspect that our measurements characterize the breakage strength of the weakened material around the weld seam and not the strength of the actual welded region. This hypothesis was reinforced by the endcap mechanical strength measurements performed on capillaries presented in the next section.
Welding the fiber and the glass slide together is an intermediate step towards the final goal, which is the assembly of endcaps to photonic crystal fibers. The next section will discuss the welding and cutting of the endcap on MOFs and capillaries, as well as a specific technique to measure the endcap’s mechanical resistance.
2.2.2 Assembling of an endcap to a microstructured optical fiber
We used the previously described method to seal off the end of a microstructured optical fiber (MOF). For this experiment, we used a MOF that was designed and drawn at our facility. This high numerical aperture MOF essentially consists of a 275 µm diameter multimode core surrounded by a ring of air-hole microstructures. Following careful polishing and cleaning, i.e. avoiding contaminating the air-holes, the MOF was welded to both 50 µm and 100 µm thick fused silica slides. Figure 9 shows microscope images of the joined area between the MOF and a 100 μm thin glass slide. A first welded region (welding region 1) was intentionally inscribed inside the core area of the fiber to solidify the bonded area. This zone aimed to seal a 100 µm diameter optical transmission window by 35 weld lines written in a spiral pattern spaced 2 µm apart. A second welded region (welding region 2) was written in the outer 25 µm thick ring between the air-hole ring and the fiber circumference. This region is composed of ten weld lines spaced by 1 µm with an inner diameter of 320 µm. Figure 9(c) shows the state of optical contact at the joined area.
The final step in the process is to mill the endcap according to the method described in section 2.1. Figure 10 shows close-up views of the edge of the endcap shown on Fig. 5. The fiber tip and endcap were doused in an isopropanol solution along with ultrasonic excitation for 20 minutes prior to observation with the SEM. This endcap was milled at a pulse energy of 700 nJ. The cross section of the cut is indeed rougher than the polished surfaces but this does not affect the endcap’s functionality. However, debris deposition is clearly visible on both the fiber and the endcap surface and it most likely originates from the ablated matter from the cutting. Note however that, because of the way the glass slide was milled, the debris was not deposited on the front surface of the endcap where light is transmitted. We observed that milling with higher energy pulses (> 1 µJ) reduces this debris deposition at the back surface, but increases the cross section roughness and the extent of the damage around the kerf, which in turn may affect the endcap mechanical strength.
Next, we estimated the endcap’s mechanical strength. In order to refine our measurement, we took advantage of the fact that the mechanical strength of endcaps deposited on airhole MOFs can be simply measured by applying gas pressure through the holes until breakage of the endcap occurs. To better control the experimental conditions, the actual MOFs were replaced by 400/200 µm outer/inner diameters fused silica capillaries. This geometry was preferred since the inner hole area is larger than that of the MOF samples, thereby reducing the air pressure required to induce breakage of the endcaps. The laser annular welding pattern was written following a spiral with an inner diameter of 240 µm composed of 30 weld lines spaced by 2 µm. Pressure was induced inside the capillary by first gluing the endcap-capillary assembly inside a steel capillary which was then attached to the exit pipes of a pressurized hydrogen bottle along with a pressure multiplier system. This setup allows the build-up of pressures inside the sealed capillary up to 4600 psi (31.7MPa). The gauge pressure measurements and corresponding tensile strength on weld at breakage for both 50 µm and 100 µm thick endcaps are presented on Fig. 11. Note that the tensile strength on weld was calculated by converting the gauge pressure to MPa (1 Psi = 6.895x10−3 MPa) and then multiplying by the ratio of the capillary’s hole area to that of the welded surface (π(1002)/(π(1802-1202)) ≈0.556).
We first note that the 50 µm thick endcaps broke at pressures ranging between 5 and 16MPa, the mean breakage pressure being at 9.2 MPa which is in good agreement with the mean values of the pull tests from Fig. 7 on standard silica fibers. However, the pressure setup was inadequate to characterise the joint strength of 100 µm thick endcaps since only one sample broke inside the applicable pressure range. The four other 100 µm endcaps resisted to the maximum available pressure. Therefore, the mean breakage pressure for this group of measurements cannot be adequately calculated but it is most certainly higher than the one obtained for 50 µm thick endcaps. This may be explained by the ability to obtain a good quality optical contact between the fiber tip and 100 μm glass slide before welding. In the case of 50 µm thick endcaps, slight lift-off of optical contact was observed inside the sealed area after welding, as it was not the case with the 100 µm endcaps. Approaching the capillary tip to a 50 µm thick glass slide to achieve optical contact will induce a slight bending of the glass slide to maintain intimate contact during welding. After milling the endcap, the glass releases this residual bending stress which may in turn induce lift-off around the welded region and trap residual mechanical constraints that may deteriorate the joint strength. This bending effect of the glass slide was less pronounced when welding the same capillaries to the 100 µm thick plates, such that very little mechanical constraints may affect the joint strength of the assemblies. Nonetheless, this test demonstrates that the endcaps assembled to capillaries and MOFs according to our method are hermetically sealed since no sign of leakage was observed until complete destruction of the endcap.
3. Femtosecond laser welding parallel to an interface
3.1 Experimental procedure
The standard approach to fs transmission laser welding with the focused beam at normal incidence with respect to the joint plane is perfectly suited for the joining of slabs or thin blocks. However, it cannot be implemented in situations where the samples to be joined are thick. A good example of this situation is the splicing of optical fibers where transmission laser welding cannot be achieved at normal incidence. Now optical fiber fusion splicing based on standard methods is very often problematic especially when dissimilar fibers (material or geometry) have to be joined. In the following we show that transmission laser welding can be transposed to situations where the laser beam is essentially parallel to the joint interface. The schematic of the proposed approach is depicted in Fig. 12. Transmission welding with fs laser pulses passing through a top substrate will simply occur as long as a part of the filament crosses the interface [7,8,10,16]. In this case the tolerance margin in positioning the laser beam is essentially fixed by the optical filament length. When considering transmission welding through the side, the alignment tolerances are very tight if one desires to expose the interface along millimeter or centimeter distances. Therefore, to enhance the positioning tolerance, samples welded through the side were slightly tilted such that the optical axis and filaments forming thereupon cross the interface at a razing angle. In our experiment, the interface is tilted at 6° which gives the user a margin of approximately 10 µm to correctly position a 100 µm long filament across the interface.
3.2 Experimental results
Rectangular fused silica substrates of 2 mm thickness were polished by Doric Lenses Inc. so as to achieve a 60-40 surface quality and a surface flatness better than λ/4. The top 10x15 mm2 windows were direct bonded to the bottom 15x20 mm2 windows. Assemblies showing interference patterns after direct bonding due to slight lift-off of optical contact were discarded. Qualified assemblies were then cut using a dicing saw (Disco 3000 series) into square substrates with 4 mm sides. The selected assemblies had their trenches polished with a 1 µm cerium oxide solution so as to obtain all sides of the block with optical quality surfaces. Proper positioning of weld lines at the interface throughout the whole length of the scan line was checked with an optical microscope and SEM after laser exposure. Figure 13 shows a SEM cross section view of the part of the elongated region of refractive index modification induced by filaments that crosses the interface at an angle of approximately 6°. Laser parameters were chosen so as to induce filaments longer than 100 µm at approximately 300 µm inside the bulk of the assemblies. Close inspection of Fig. 13 reveals that the faint line defining the separation between materials is erased along the modified region.
In order to confirm that successful welding and reinforcement has occurred, we measured the joint strength of assemblies with and without weld lines. Shear strength testing conditions and setup are the same as those from . We prepared 45 square fused silica assemblies with 4 mm sides which were separated into three groups of 15 samples. The first was not exposed to the laser and served as a reference for the strength of the direct bond. The second group was processed through the side by writing one 3.7 mm long weld line along each edge so as to seal the central region. The third group was exposed through the top substrate along each side such that the optical filaments cross the interface at normal incidence. The sealed area and laser parameters for groups 2 and 3 are identical except for the tilt of the assemblies. Figure 14 shows a Weibull plot of the shear strength measurements at breakage of the bonded assemblies (dots) where each set of measurements is fitted to the two-parameter Weibull model (solid line) . The error bars were adjusted according to the maximum difference between the values from the model and those from the measurements. Strength is derived from the whole 4x4 mm2 direct bonded area.
The Weibull fits statistically assess the likelihood that a bonded sample will break following the application of a specific amount of shear strain. Results clearly show that the assemblies containing weld lines exhibit a greater resistance to applied shear loads than those unprocessed. As a side benefit, the optical transmission of assemblies inside the sealed area is unaltered by the welding process. Components reinforced by weld lines written through the side show a slightly smaller resistance to those reinforced through the top. Nevertheless, a mean reinforcement factor superior to three is obtained for both configurations after welding less than 1% of the direct bonded surface. This is accounted for by the mixing of material and rapid transformation of Van der Waals bonds into covalent bonds throughout the weld lines. The intrinsic strength and bond energy of such welded regions has typically been reported to be one to two orders of magnitude greater than that of a direct bonded surface [20,21]. These results demonstrate the versatility and potential of fs laser microwelding for application in different geometrical configurations. For instance, it should be possible based on the preceding preliminary results to transpose this method to a cylindrical geometry to apply it to the splicing of optical fibers.
We have demonstrated the potential of femtosecond laser welding for the reinforcement of the direct bond between a fiber tip and glass slides. Based on a statistical analysis we have shown that the resulting assemblies typically possess tensile stress at breakage of the order of 10 MPa but could be significantly enhanced by improving the conditions of optical contact between the surfaces. Based on this joining approach, we have demonstrated that optical fiber endcaps could be fabricated in a very efficient and convenient way by the use of the same laser system as a milling tool. This concept was demonstrated for standard fibers, MOFs as well as capillaries. In the case of MOFs, it was shown that hermetic seals could be achieved without damaging the air-hole microstructures. Finally we have presented preliminary results on femtosecond laser welding parallel to an interface showing the potential of this approach for optical fiber fusion splicing. These results are paving the way to a novel approach to the fabrication of optical fiber endcaps as well as fusion splicing that would be free of core deformation and would therefore minimally affect fiber mode transmission.
The authors would like to thank the manufacturing team at Doric Lenses, Inc., for their help in defining the proper fiber preparation strategy. Special thanks to Steeve Morency and Stéphan Gagnon for sample preparation and characterisation. Financial support from the FQRNT (Fonds de Recherche du Québec-Nature et Technologies), the NSERC (National Sciences and Engineering Research Council of Canada), the Canada Foundation for Innovation (CFI) is also acknowledged.
References and links
1. P. Russell, “Photonic-crystal fibers,” J. Lightwave Technol. 24(12), 4729–4749 (2006). [CrossRef]
2. Y. Ariel, E. Sherman, A. Patlakh, and R. Bronstein, “Termination of end-faces of air-clad and photonic-crystal fibers,” U.S. patent 2003/0068150 (10 April, 2003).
3. A. D. Yablon, Optical Fiber Fusion Splicing (Springer, Heidelberg, 2005).
4. S. Böhme, S. Fabian, T. Schreiber, R. Eberhardt, and A. Tünnermann, “End cap splicing of photonic crystal fibers with outstanding quality for high power applications,” Proc. SPIE 8244, 824406, 824406-9 (2012). [CrossRef]
5. S. Boehme, E. Beckert, R. Eberhardt, and A. Tuennermann, “Laser splicing of end caps – process requirements in high power laser applications,” Proc. SPIE 7202, 720205, 720205-11 (2009). [CrossRef]
6. S. Sinha, K. E. Urbanek, A. Krzywicki, and R. L. Byer, “Investigation of the suitability of silicate bonding for facet termination in active fiber devices,” Opt. Express 15(20), 13003–13022 (2007). [CrossRef] [PubMed]
7. T. Tamaki, W. Watanabe, J. Nishii, and K. Itoh, “Welding of transparent materials using femtosecond laser pulses,” Jpn. J. Appl. Phys. 44(22), L687–L689 (2005). [CrossRef]
9. S. Richter, S. Döring, A. Tünnermann, and S. Nolte, “Bonding of glass with femtosecond laser pulses at high repetition rates,” Appl. Phys., A Mater. Sci. Process. 103(2), 257–261 (2011). [CrossRef]
10. J. Nishii, W. Watanabe, S. Onda, T. Tamaki, and K. Itoh, “Space-selective laser joining of dissimilar transparent materials using femtosecond laser pulses,” Appl. Phys. Lett. 89(2), 021106 (2006). [CrossRef]
11. C. B. Schaffer, A. Brodeur, and E. Mazur, “Laser-induced breakdown and damage in bluk transparent materials induced by tightly focused femtosecond laser pulses,” Meas. Sci. Technol. 12(11), 1784–1794 (2001). [CrossRef]
12. J. Haisma and G. A. C. M. Spierings, “Contact bonding, including direct-bonding in a historical and recent context of materials science and technology, physics and chemistry,” Mater. Sci. Eng. Rep. 37(1-2), 1–60 (2002). [CrossRef]
13. R. R. Gattass and E. Mazur, “Femtosecond laser micromachining in transparent materials,” Nat. Photonics 2(4), 219–225 (2008). [CrossRef]
14. A. Couairon and A. Mysyrowicz, “Femtosecond filamentation in transparent solids,” Phys. Rep. 441(2-4), 47–189 (2007). [CrossRef]
15. B. Poumellec, M. Lancry, A. Chahid-Erraji, and P. G. Kazansky, “Modification thresholds in femtosecond laser processing of pure silica: review of dependencies on laser parameters,” Opt. Mater. Express 1(4), 766–782 (2011). [CrossRef]
16. D. Hélie, F. Lacroix, and R. Vallée, “Reinforcing a direct bond between optical materials by filamentation based femtosecond laser welding,” Journ. Las. Micr/Nano Eng. JLMN 7(3), 284–292 (2012). [CrossRef]
17. D. J. Huang, T. Y. Choi, and C. P. Grigoropoulos, “Liquid-assisted femtosecond laser drilling of straight and three-dimensional microchannels in glass,” Appl. Phys., A Mater. Sci. Process. 79(3), 605–612 (2004). [CrossRef]
18. X. Zhao and Y. C. Shin, “Femtosecond laser drilling of high-aspect ratio microchannels in glass,” Appl. Phys., A Mater. Sci. Process. 104(2), 713–719 (2011). [CrossRef]
19. R. B. Abernethy, The New Weibull Handbook (R.B. Abernethy, Florida, 2000).
20. K. Cvecek, I. Miyamoto, J. Strauss, M. Wolf, T. Frick, and M. Schmidt, “Sample preparation method for glass welding by ultrashort laser pulses yields higher seam strength,” Appl. Opt. 50(13), 1941–1944 (2011). [CrossRef] [PubMed]
21. S. Roth, K. Cvecek, I. Miyamoto, and M. Schmidt, “Glass welding technology using ultra short laser pulses,” Proc. SPIE 7920, 7920006 (2011). [CrossRef]