Characterization and mechanism of glass ... - OSA Publishing

Characterization and mechanism of glass microwelding by double-pulse ultrafast laser irradiation Sizhu Wu,1,2 Dong Wu,1 Jian Xu,1 Yasutaka Hanada,1 Ryo Suganuma,1 Haiyu Wang,2 Testuya Makimura,3 Koji Sugioka,1* and Katsumi Midorikawa1 1

Laser Technology Laboratory, RIKEN – Advanced Science Institute, 2-1 Hirosawa, Wako, Saitama 351-0198, Japan 2 State Key Laboratory on Integrated Opto-electronics, College of Electronic Science and Engineering, Jilin University, 2699 Qianjin Road, Changchun 130012, China 3 Institute of Applied Physics, University of Tsukuba, 1-1-1 Ten’nodai, Tsukuba 305-8573, Japan *[email protected]

Abstract: We investigated the physical mechanism of high-efficiency glass microwelding by double-pulse ultrafast laser irradiation by measuring the dependences of the size of the heat-affected zone and the bonding strength on the delay time between the two pulses for delay time up to 80 ns. The size of the heat-affected zone increases rapidly when the delay time is increased from 0 to 12.5 ps. It then decreases dramatically when the delay time is further increased to 30 ps. It has a small peak around 100 ps. For delay time up to 40 ns, the size of the heat-affected zone exceeds that for a delay time of 0 ps, whereas for delay time over 60 ps, it becomes smaller than that for a delay time of 0 ps. The bonding strength exhibits the same tendency. The underlying physical mechanism is discussed in terms of initial electron excitation by the first pulse and subsequent excitation by the second pulse: specifically, the first pulse induces multiphoton ionization or tunneling ionization, while the second pulse induces electron heating or avalanche ionization or the second pulse is absorbed by the localized state. Transient absorption of glass induced by the ultrafast laser pulse was analyzed by an ultrafast pump–probe technique. We found that the optimum pulse energy ratio is unity. These results provide new insights into highefficiency ultrafast laser microwelding of glass and suggest new possibilities for further development of other ultrafast laser processing techniques. ©2012 Optical Society of America OCIS codes: (140.3390) Laser materials processing; (140.7090) Ultrafast lasers; (160.2750) Glass and other amorphous materials; (190.4180) Multiphoton processes.

References and links 1. 2. 3. 4. 5. 6. 7.

T. Tamaki, W. Watanabe, J. Nishii, and K. Itoh, “Welding of transparent materials using femtosecond laser pulse,” Jpn. J. Appl. Phys. 44(22), L687–L689 (2005). W. Watanabe, S. Onda, T. Tamaki, K. Itoh, and J. Nishii, “Space-selective laser joining of dissimilar transparent materials using femtosecond laser pulse,” Appl. Phys. Lett. 89(2), 021106 (2006). T. Tamaki, W. Watanabe, H. Nagai, M. Yoshida, J. Nishii, and K. Itoh, “Structural modification in fused silica by a femtosecond fiber laser at 1558 nm,” Opt. Express 14(15), 6971–6980 (2006). C. Luo and L. Lin, “The application of nanosecond-pulsed laser welding technology in MEMS packaging with a shadow mask,” Sens. Actuators A Phys. 97-98, 398–404 (2002). W. Watanabe, S. Onda, T. Tamaki, and K. Itoh, “Direct joining of glass substrates by 1 kHz femtosecond laser pulses,” Appl. Phys. B 87(1), 85–89 (2007). A. Horn, I. Mingareev, A. Werth, M. Kachel, and U. Brenk, “Investigations on ultrafast welding of glass-glass and glass-silicon,” Appl. Phys., A Mater. Sci. Process. 93(1), 171–175 (2008). Y. Ozeki, T. Inoue, T. Tamaki, H. Yamaguchi, S. Onda, W. Watanabe, T. Sano, S. Nishiuchi, A. Hirose, and K. Itoh, “Direct welding between copper and glass substrates with femtosecond laser pulses,” Appl. Phys. Express 1, 082601 (2008).

#177531 - $15.00 USD

(C) 2012 OSA

Received 5 Oct 2012; revised 15 Nov 2012; accepted 17 Nov 2012; published 12 Dec 2012

17 December 2012 / Vol. 20, No. 27 / OPTICS EXPRESS 28893

8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25.

Y. Kim, J. Choi, Y. Lee, T. Kim, D. Kim, W. Jang, K.-S. Lim, I.-B. Sohn, and J. Lee, “Femtosecond laser bonding of glasses and ion migration in the interface,” Appl. Phys., A Mater. Sci. Process. 101(1), 147–152 (2010). I. Miyamoto, K. Cvecek, Y. Okamoto, M. Schmidt, and H. Helvajian, “Characteristics of laser absorption and welding in FOTURAN glass by ultrashort laser pulses,” Opt. Express 19(23), 22961–22973 (2011). I. Miyamoto, K. Cvecek, and M. Schmidt, “Evaluation of nonlinear absorptivity in internal modification of bulk glass by ultrashort laser pulses,” Opt. Express 19(11), 10714–10727 (2011). I. Miyamoto, K. Cvecek, and M. Schmidt, “Evaluation of nonlinear absorptivity and absorption region in fusion welding of glass using ultrashort laser pulse,” Phys. Procedia 12, 378–386 (2011). S. Richter, S. Doring, A. Tunnermann, 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). D. Hélie, M. Bégin, F. Lacroix, and R. Vallée, “Reinforced direct bonding of optical materials by femtosecond laser welding,” Appl. Opt. 51(12), 2098–2106 (2012). K. Sugioka, M. Iida, H. Takai, and K. Micorikawa, “Efficient microwelding of glass substrates by ultrafast laser irradiation using a double-pulse train,” Opt. Lett. 36(14), 2734–2736 (2011). S. M. Eaton, H. Zhang, M. L. Ng, J. Li, W. J. Chen, S. Ho, and P. R. Herman, “Transition from thermal diffusion to heat accumulation in high repetition rate femtosecond laser writing of buried optical waveguides,” Opt. Express 16(13), 9443–9458 (2008). C. Y. Ho, “Effects of polarizations of a laser on absorption in a paraboloid of revolution-shaped welding or drilling cavity,” J. Appl. Phys. 96(10), 5393–5401 (2004). S. M. Eaton, H. Zhang, P. Herman, F. Yoshino, L. Shah, J. Bovatsek, and A. Arai, “Heat accumulation effects in femtosecond laser-written waveguides with variable repetition rate,” Opt. Express 13(12), 4708–4716 (2005). M. Shimizu, M. Sakakura, M. Ohnishi, Y. Shimotsuma, T. Nakaya, K. Miura, and K. Hirao, “Mechanism of heat-modification inside a glass after irradiation with high-repetition rate femtosecond laser pulses,” J. Appl. Phys. 108(7), 073533 (2010). C. B. Schaffer, J. F. Garcia, and E. Mazur, “Bulk heating of transparent materials using a high-repetition-rate femtosecond laser,” Appl. Phys., A Mater. Sci. Process. 76(3), 351–354 (2003). S. S. Mao, F. Quere, S. Guizard, X. Mao, R. E. Russo, G. Petite, and P. Martin, “Dynamics of femtosecond laser interactions with dielectrics,” Appl. Phys., A Mater. Sci. Process. 79, 1695–1709 (2004). R. R. Gattass and E. Mazur, “Femtosecond laser micromachining in transparent materials,” Nat. Photonics 2(4), 219–225 (2008). S. Guizard, P. D’Oliveira, P. Daguzan, P. Martin, P. Meynadier, and G. Petite, “Time-resolved studies of carriers dynamics in wide band gap materials, ” Nucl. Instr. and Meth. in Phys. Res. B 116, 43–48 (1996). B. C. Stuart, M. D. Feit, S. Herman, A. M. Rubenchik, B. W. Shore, and M. D. Perry, “Nanosecond-tofemtosecond laser-induced breakdown in dielectrics,” Phys. Rev. B Condens. Matter 53(4), 1749–1761 (1996). T. Hongo, K. Sugioka, H. Niino, Y. Cheng, M. Masuda, I. Miyamoto, H. Takai, and K. Midorikawa, “Investigation of photoreaction mechanism of photosensitive glass by femtosecond laser,” J. Appl. Phys. 97(6), 063517 (2005). K. S. Song and R. T. Williams, Self-Trapped Excitons (Springer-Verlag, 1993).

1. Introduction In recent years, glass microwelding has attracted great interest because of its potential application in fields such as microelectromechanical systems, precision machinery, healthcare, and small satellites. Due to the extremely high peak intensities they generate, ultrafast lasers can be used to perform rapid, high-precision, high-quality, and flexible welding of glass [1–3]. Unlike conventional laser pulses used for laser microwelding that suffer from low absorption in transparent glass materials [4], ultrafast laser pulses are strongly absorbed by such materials due to nonlinear processes such as multiphoton absorption and tunneling ionization. Consequently, they do not require intermediate layers to be inserted when they are used for glass welding. They induce local melting and rapid resolidification at the interface between two glass substrates. Several research groups have demonstrated microwelding of glass substrates using ultrafast laser pulses [1–3, 5–10]. For example, Tamaki et al. [1] used near-infrared femtosecond laser pulses to microweld two pieces of transparent silica without using an intermediate layer (such as an adhesive) for the first time. Material near the focal point of a focused femtosecond laser beam can be melted and resolidified by heating induced by localized nonlinear absorption of the optical pulse energy [10, 11]. Watanabe et al. [2] joined borosilicate glass and fused silica, which have different coefficients of thermal expansion. Horn et al. [6] systematically investigated the melting and welding properties of glass induced by femtosecond laser irradiation. To enhance the bonding

#177531 - $15.00 USD

(C) 2012 OSA



strength, optical contacting by Van der Waals forces has been employed as a pre-joining method prior to welding [9–13]. As an alternative method to enhance the efficiency of glass microwelding, we recently proposed a new strategy that involves irradiating a two-pulse train with a separation of 10 ps between the pulses. This technique increased the bonding strength of photosensitive glass welding by approximately 22% relative to that for conventional single-pulse-train irradiation [14]. This enhanced bonding strength is thought to be due to selective control of electron excitation processes by the two pulses: specifically, the first pulse induces multiphoton ionization or tunneling ionization and the second pulse induces electron heating or avalanche ionization. Although double-pulse irradiation greatly improves glass welding, its underlying physical mechanism is currently unclear. Additionally, our previous study employed delay time between the two pulses only in the range 0–10 ps. It is thus essential to systematically investigate glass microwelding by ultrafast laser double-pulse trains under various conditions both to optimize the microwelding process and to determine its physical mechanism. In this study, we extend the delay time to 80 ns and investigate the dependences of the size of the heat-affected zone and the bonding strength on the delay time. To discuss the underlying physical mechanism, an ultrafast pump–probe technique is used to investigate transient absorption of glass irradiated by ultrafast laser pulses. We also determine the optimum energy ratio for the two pulses for high-efficiency glass welding. 2. Experimental setup

Fig. 1. (a) Schematic illustration of experimental setup used for microwelding photosensitive glass substrates by irradiation of double-pulse train. (b) Irradiation scheme for evaluation of the heat-affected zone. (c) Definition of the heat-affected zone based on optical microscopy observation.

Figure 1(a) shows a schematic illustration of the experimental setup used for welding glass substrates by double-pulse irradiation. An amplified femtosecond Er-fiber laser system (IMRA America, FCPA μJewel D-400) generated 360-fs pulses with a wavelength of 1045 nm at a repetition rate of 200 kHz. The linearly polarized femtosecond laser pulses were transformed into s- and p-polarized pulses with different pulse energy ratios by rotating halfwave plate 1; these pulses were then split by a polarized beam splitter (PBS 1). The total pulse energy of the p- and s-polarized pulses in front of the objective lens was set to 1.55 μJ. The delay time was controlled by adjusting the optical path in the optical delay circuit using a high-precision stage. The substrates used were made from a commercially available photosensitive glass (Foturan; Schott Glass Corp.) that consists of lithium aluminosilicate #177531 - $15.00 USD

(C) 2012 OSA



doped with trace amounts of silver, cerium, sodium, and antimony. To evaluate the heataffected zones produced by laser irradiation, femtosecond laser pulses were focused at the same position 60 μm below the glass surface for a total irradiation time of 0.2 s by a × 20 objective lens (Mitutoyo, M Plan Apo NIR) with a numerical aperture of 0.4 [Fig. 1(b)]. The irradiated regions were then observed from both the top and side surfaces by an optical transmission microscope (Olympus, BX51) with white-light illumination to evaluate the heataffected zone. The laser-irradiated regions exhibited dark inner areas and light outer areas, as schematically illustrated in Fig. 1(c). The size of the heat-affected zone was defined as the diameter of the light outer area.

Fig. 2. (a) Schematics of microwelding of two glass substrates by focused femtosecond laser. (b) Scanning scheme in x-y plane. (c) Schematic diagram of tensile tester.

Prior to laser welding, two glass substrates were carefully cleaned by acetone and ethanol for several minutes. They were then closely stacked and pressed together by a fixture using a lens with three bolts to eliminate the air gap between them to prevent laser ablation at the interface [3]. The contact area between the two glass substrates was approximately 10 mm × 10 mm. This stacking procedure did not produce optical contact since this study used asreceived photosensitive glass substrates that did not have a sufficiently high surface smoothness and flatness to realize optical contact. The sample was mounted on a threedimensional translation stage that had a resolution of 1 μm (SK140-100). The laser beam was focused 100 μm below the interface between the two glass substrates [Fig. 2(a)] and then scanned over an area of 1 mm × 1 mm in the x–y plane at a scanning speed of 200 μm/s using the scanning scheme shown in Fig. 2(b). After scanning, the focused laser beam was translated upward parallel to the laser incident direction (z-direction) by 30 μm and it was again scanned in the x–y plane using the same scanning scheme. This procedure was repeated seven times (the shift in the z-direction before the seventh scan was 20 μm); thus, the focused laser beam was shifted by a total of 200 μm in the z-direction. In principle, single-layer scanning at the interface between the stacked two substrates is sufficient for welding. However, to ensure the interface melted over the whole irradiated area, we employed this seven-layer scanning scheme since the photosensitive glass substrates used did not have a sufficiently high smoothness and flatness. To optimize the laser microwelding process and to investigate its physical mechanism, various parameters of the double-pulse irradiation of ultrafast laser were varied. To evaluate the bonding strength, a conventional tensile tester was used [Fig. 2(c)] to pull the welded glass substrates perpendicular to the welding plane.

#177531 - $15.00 USD

(C) 2012 OSA



3. Optimization of process parameters 3.1 Delay-time dependence of heat-affected zone

Fig. 3. Optical microscope images of laser irradiated regions for delay time ranging from 0 to 40 ns and different irradiation times ranging from 0.1 to 10 s.

Delay time between 0 and 80 ns and irradiation times between 0.1 and 10 s were employed to investigate double-pulse ultrafast laser irradiation. Both pulses have an energy of 0.775 μJ. The optical microscopy images in Fig. 3 reveal that the laser-modified regions consist of dark inner areas and light outer areas. The heat-affected zone is considered to be made up of both regions. The dark inner region where the refractive index is increased is directly produced by the high temperature and high pressure generated by the laser pulse. Increasing the irradiation time expands the high-temperature and high-pressure region and thus increases the diameter of the dark inner area. This diameter also depends significantly on the delay time. The heat generated in the dark inner region can diffuse to the surroundings and melt regions, which are much larger than the focal volume. This heat diffusion produces the light outer regions that have a lower refractive index than the inner dark regions. Such a core–cladding structure is often produced by high-repetition-rate femtosecond lasers [15]. This melting enables fusion welding to be performed. As seen from Fig. 3, the size of the light outer region also increases as the irradiation time increases due to the increased dosage. In addition, delay time longer than 15 ps clearly produce larger heat-affected zones than when there is no delay between the two pulses (i.e., the delay time is 0 ps). However, the heat-affected zone becomes smaller when the delay time is increased beyond 15 ps, which suggests that there is an optimal delay time for microwelding. To determine the optimal delay time for high-efficiency microwelding, the relationship between the delay time and the size of the heat-affected zone was quantitatively investigated, as shown in Fig. 4(a). Here, the size of the heat-affected zone is defined as the diameter of the light outer region [Fig. 1(c)]. The size of the heat-affected zone increased rapidly as the delay time was increased to 12.5 ps; the maximum heat-affected zone size of 33.46 μm was obtained at a delay time of 12.5 ps. The size of the heat-affected zone decreased significantly as the delay time was increased from 15 to 30 ps (where its size was 31.45 μm). There was a

#177531 - $15.00 USD

(C) 2012 OSA



small peak around 100 ps. The size of the heat-affected zone decreased to 26.21 μm at a delay time of 40 ns, but this was still larger than the heat-affected zone produced when there was no delay between the two pulses (i.e., the delay time was 0 ps).

Fig. 4. (a) Dependence of size of the heat-affected zone on delay time. The first and second pulses both have a duration of 0.2 s and a pulse energy of 0.775 μJ. The red circle and blue triangle indicate the sizes of the heat-affected zones produced by a conventional single-pulse train (pulse energy: 1.55 μJ) for p- and s-polarized beams, respectively. (b) Optical microscopy images for delay times of 0 ps, 60 ns, and 80 ns (pulse energy: 1.35 μJ).

Since the delay time at which the size of the heat-affected zone becomes smaller than that for no delay gives important insight into the mechanism, investigation at longer delay time is necessary. However, such delay time require optical delay paths longer than a few meters, which result in significant optical loss. Since our laser system has a maximum pulse energy of 1.55 μJ (which is the total pulse energy employed in the above experiments), samples were prepared with a total pulse energy of 1.35 μJ (i.e., 0.675 μJ per pulse) for delay time longer than 40 ns and the results were compared with those obtained for the same total pulse energy with no delay; these results are shown in Fig. 4(b). The sizes of the heat-affected zones at delay time of 60 and 80 ns are respectively estimated to be 8.96 and 6.42 μm, which are smaller than that (13.62 μm) produced with no delay using the same total pulse energy. This implies that samples prepared using delay time longer than 60 ns will have inferior welding characteristics to those prepared using no delay time. In fact, welding could be performed for a total pulse energy of 1.35 μJ when there was no delay, whereas no welding occurred for delay time longer than 60 ps.

#177531 - $15.00 USD

(C) 2012 OSA



Fig. 5. (a) Cross-sectional images of laser-irradiated regions for delay times between 0 and 30 ps and exposure times between 0.1 and 10 s. (b) Dependence of the vertical length of the crosssection of heat-affected zone on exposure time for different delay time.

The sizes of the heat-affected zones of samples prepared by conventional irradiation with a single-pulse train were also evaluated. The results are shown in Fig. 4(a), where the red circle and blue triangle indicate the results for p- and s-polarized beams, respectively. The incident total pulse energy was the same in all cases; namely, the single-pulse train had a pulse energy of 1.55 μJ, whereas the double-pulse train consisted of two pulses with a pulse energy of 0.775 μJ. Interestingly, single-pulse train irradiation produced a larger heat-affected zone than simultaneous irradiation (i.e., delay time: 0 ps) of p- and s-polarized beams. This is because multiphoton ionization or tunneling ionization occurs independently for a beam with a specific polarization so that p- and s-polarized beams cannot cooperatively contribute to these ionization processes to excite a single photon. Therefore, double-pulse irradiation at a delay time of 0 ps cannot efficiently excite electrons from the valence band to the conduction band. In single-pulse train irradiation, the p-polarized beam produces a larger heat-affected zone than the s-polarized beam. This is probably due to higher reflection of the s-polarized beam at the interface between the heat-affected zone and bulk glass. The heat-affected zone has an elliptical cross-section [Fig. 5(a)] so that the s-polarized beam should be reflected more at this curved interface due to the refractive index difference between the heat-affected

#177531 - $15.00 USD

(C) 2012 OSA



zone and bulk glass. A similar reduction in the processing efficiency of laser material processing is often observed when using an s-polarized beam [16]. The vertical length of the cross-sections of the irradiated regions was also examined. Fig. 5(a) shows cross-sectional optical microscopy images of laser-irradiated regions of samples prepared by double-pulse irradiation using different delays and irradiation times. The modified structures are elliptical and their longitudinal dimensions are clearly larger than their axial dimensions. This is responsible for the mismatch between the focal radius and the Rayleigh length of the focused laser beam. The size of the laser-affected zone of the crosssections increased with increasing irradiation time, which is consistent with the variation in the x–y plane shown in Figs. 3 and 4. This increase is probably due to the heat accumulation effect. Typically, heat accumulation occurs at repetition rates greater than a few hundred kHz, although it depends on the kind of glass and the fluence [15, 17–19]. The maximum size of the laser-affected zone was obtained at a delay time of about 15 ps [Fig. 5(b)]. 3.2 Delay-time dependence of bonding strength

Fig. 6. Dependence of bonding strength on delay time. The bonding strength exhibits the same tendency as the size of the heat-affected zone. It increases rapidly when the delay time is increased from 0 to 12.5 ps, but it decreases dramatically above 12.5 ps. It almost saturates between 30 ps and 2 ns with a small peak at 100 ps and it decreases gradually in the range 1–2 ns.

The bonding strength of welded samples was also investigated. Welding was performed by scanning the focused laser beam on the interface between two tightly stacked glass substrates [Fig. 2(b)] for different delay time. Figure 6 shows the relationship between the bonding strength and the delay time. The bonding strength increased rapidly from 10.52 to 13.36 MPa (27% increase) when the delay time was increased from 0 to 15 ps. However, the bonding strength decreased abruptly when the delay time was increased from 15 ps to about 30 ps. It showed a small peak around 100 ps and then decreased gradually. The bonding strengths of 11–11.5 MPa obtained for delays between 30 ps and 40 ns is still slightly greater than that obtained for a delay time of 0 ps and for single-pulse irradiation. Thus, the bonding strength exhibits a similar tendency to the size of the heat-affected zone. Single-pulse irradiation of pand s-polarized beams and double-pulse irradiation with a delay time of 0 ps show the same correlation as the heat-affected zone. For this welding condition, the width of the heataffected zone was measured to be about 16 μm at a delay time of 12.5 ps, which is narrower

#177531 - $15.00 USD

(C) 2012 OSA



than the distance of 30 μm between adjacent lateral scanning lines. This clearly indicates that the bonding strength is strongly associated with the lateral size of the heat-affected zone. The bonding strength obtained in our experiments was much smaller than that (~100 MPa) achieved for photosensitive glass by Miyamoto et al. [9]. This is due to the different preparation conditions used, such as the repetition rate, the scanning speed, and the pulse energy. In particular, a higher repetition rate enhances larger heat accumulation, which produces a larger heat-affected zone. Moreover, Miyamoto et al. used sample pairs with optical contact, which greatly increases the bonding strength in laser welding [1]; in contrast, optical contact was not realized in the present study. The bonding strength achieved for borosilicate glass is comparable with that obtained in this study [5]. 3.3 Energy-ratio dependence of bonding strength

Fig. 7. Dependence of bonding strength on ratio of energies of first and second pulses. The optimal energy ratio is unity.

In Secs. 3.1 and 3.2, both pulses had the same pulse energy. To optimize the processing parameters, dependence of the bonding strength on the energy ratio of the two pulses for a total pulse energy of 1.55 μJ was investigated. The results are shown in Fig. 7. The energy ratio is calculated by dividing the first pulse energy (p-polarized beam) by the second pulse energy (s-polarized beam). Double-pulse trains were irradiated using the same scheme as that employed in Sec. 3.2 for a delay time of 15 ps and the energy ratio was varied between 0.25 and 5. The results obtained clearly indicate that the bonding strength is maximized when this ratio is unity (i.e., the two pulses have the same pulse energy). 4. Discussion of physical mechanism This section discusses the underlying physical mechanism of ultrafast laser microwelding by double-pulse irradiation based on the above experimental results. Glass welding by ultrafast laser irradiation is considered to occur due to melting induced at the interface between two glass substrates by irradiation of a focused laser beam. The electron excitation and relaxation processes in glass induced by ultrafast laser irradiation are as follows [20, 21]. Electrons are first excited from the valence band to the conduction band on a timescale of a few hundreds of femtoseconds by multiphoton absorption of the ultrafast laser light (multiphoton ionization) or tunneling ionization when the electromagnetic field of the laser is extremely strong. The excited electrons can successively absorb several laser photons and be excited to higher energy states where free carrier absorption is efficient (electron heating). In addition, #177531 - $15.00 USD

(C) 2012 OSA



when the laser intensity is sufficiently high, the excited electrons are accelerated by the intense electric field of the ultrafast laser beam and they collide with surrounding atoms, generating secondary electrons (avalanche ionization). Generated free electrons relax to localize the energy stored in electron–hole pairs, which creates self-trapped excitons (STEs). This relaxation often starts at times shorter than 1 ps after laser irradiation. Some STEs relax to form permanent defects on a timescale between a few hundred picoseconds and a few nanoseconds. Within a couple of nanoseconds, a pressure or shock wave separates from the dense, hot focal volume. Glass heating also occurs a few tens of picoseconds after laser irradiation and the thermal energy diffuses out of the focal volume on a timescale of microseconds. Finally, the irradiated area returns to room temperature after several tens of micrometers, resulting in damage formation. Melting due to heating was observed when an ultrafast laser beam was focused inside glass [17, 20]. Control of electron excitation will be important for efficient heating that produces a larger melted pool for efficient and highquality welding. Although the processes that occur after electron excitation in glass by ultrafast laser irradiation are complex since they involve intermediate processes such as plasma formation, plasma relaxation, thermalization, and thermal diffusion, depositing as high a laser energy as possible into the glass is the most important factor for realizing efficient heating. Therefore, we discuss the mechanism of glass welding by double-pulse irradiation below based on transient absorption changes induced by electron excitation and relaxation.

Fig. 8. A possible physical mechanism of glass welding by double-pulse irradiation. It involves electron excitation and relaxation processes including multiphoton ionization or tunneling ionization, avalanche ionization or electron heating, and the electron trapping at the localized state.

Figure 8 summarizes the electron excitation and relaxation processes in glass that are speculated to be involved in the microwelding mechanism. The first pulse excites free electrons (i.e., seed electrons) from the valance band to the conduction band by multiphoton absorption or tunneling ionization within a few hundred femtoseconds [Fig. 8(a) depicts multiphoton ionization]. These free electrons couple with the lattice over a time scale ranging from less than a picosecond to several tens of picoseconds depending on the material [22, 23]. If the second pulse is irradiated during this time scale, its energy can be absorbed by the excited electrons. The electrons will then be excited to higher energy levels so that electron heating occurs [Fig. 8(b)]. Additionally, if the laser intensity is sufficiently high, the energy of the heated electrons will exceed the conduction band minimum by more than the bandgap energy. It can thus ionize another electron from the valence band, resulting in two excited

#177531 - $15.00 USD

(C) 2012 OSA



electrons at the conduction band minimum (impact ionization). This process is repeated by the intense electric field of the second pulse and eventually free electrons are multiplied (avalanche ionization) [Fig. 8(c)]. These processes are induced successively by the two laser pulses (i.e., multiphoton absorption or tunneling ionization induced by the first pulse followed by electron heating or avalanche ionization induced by the second pulse) and can generate higher-energy free electrons or more free electrons for efficient heating [14], which imparts a higher bonding strength for welding. The highest bonding strength was obtained for delay time in the range 12.5–15 ps. Our previous study [14] and this study reveal that both the bonding strength and the size of the heat-affected zone increase rapidly as the delay time is increased up to 2 ps and they subsequently become almost saturated at a delay time of 15 ps. The increase up to 2 ps is probably due to the first pulse generating less than the optimal amount of seed electrons before the second pulse arrives since interband excitation of electrons in photosensitive glass by an ultrafast laser beam is relatively complex (i.e., successive interband electron excitation through defect levels by multiphoton absorption) [24]. This may cause the excitation rate to be longer than hundreds of femtoseconds in typical glasses. The time range in which the bonding strength is maximized (2–15 ps) is much longer than the relaxation time of free electrons in fused silica (less than 0.5 ps) [22]; this may be due to photosensitive glass having a more complex system than fused silica. On the other hand, if the second laser pulse is irradiated a couple of tens of picoseconds after the first pulse, the second pulse will not interact with the free electrons since they have relaxed and consequently the bonding strength will be dramatically reduced. Some excited electrons will relax to the valence band, generating heat, while other electrons may be trapped in a localized state that has a longer relaxation time due to defects or excitons. For example, SETs in fused silica have a very long lifetime (typically longer than 1 ns) [25]. The second pulse can still be absorbed by such a localized state to create an excited state even for delay time longer than a couple of tens of picoseconds [Fig. 8(d)]. It is currently not known whether a single photon induces this absorption. However, even for multiphoton absorption, it may be more efficiently induced than interband excitation shown in Fig. 8(a) since the order of multiphoton absorption should be lower than that for interband excitation. Excitation from the localized state to the conduction band is another possible channel for this absorption, rather than the formation of an excited state. Electrons excited from the localized state eventually relax to the ground state or the valence band, generating heat. This absorption of the localized state by the second pulse explains why double-pulse irradiation with delay time between 30 ps and 40 ns gives a higher bonding strength than single-pulse irradiation. After several tens of nanoseconds, electrons trapped in the localized state will relax and absorption of the localized state will no longer occur. Consequently, for delay time longer than 60 ns, both the bonding strength and the size of the heat-affected zone will be smaller than those for single-pulse irradiation. Heat generated by the first pulse may be another factor that increases the absorption of the second pulse. However, it takes several tens of microseconds for the heated area to return to room temperature. Therefore, absorption by the localized state is more likely based on the timescale of the transient absorption change (see Fig. 9). It remains to explain the small peak observed at about 100 ps for both the bonding strength and the size of the heat-affected zone. Electrons are expected to relax from the conduction band to the localized state in a time of about 100 ps. However, the relaxation rate is unlikely to be very much longer than that from the conduction band to the valence band [a couple of tens of picoseconds; Fig. 8(a)]. Thus, there may be another state that has a relaxation time of several tens of picoseconds. Electrons may relax from the conduction band to the localized state via this intermediate state, which would explain the present observations. Further investigation is necessary to confirm this conjecture.

#177531 - $15.00 USD

(C) 2012 OSA



Fig. 9. Transient absorption of second pulse for various delay times.. The transmittance decreases from 0 to 15 ps, and then increases from 15ps. It has a small dip at 100 ps.

To strengthen the above discussion, the transient absorption of photosensitive glass induced by ultrafast laser pulse irradiation was evaluated. The experimental scheme used for this measurement is the same as that described above. A PBS and a power meter were used to measure the energy of the second pulse (s-polarized beam) transmitted through the glass (see inset of Fig. 9) for different delay time. In this experiment, the pulse energies were the same as those used in Figs. 4 and 6 to measure absorption during welding. Figure 9 shows that the transmittance is maximized at a delay time of 0 ps (49.37%). As the delay time was increased, the transmittance decreased rapidly to a minimum value of 41.05% at a delay time of 12.5 ps. It then increased dramatically to 44.21% as the delay time was increased to 30 ps. It exhibits no large variation for delay time between 30 ps and a couple tens of nanoseconds, although it has a small dip at around 100 ps. In another experiment, we confirmed that the reflection of second pulse was almost the same for all delay time. Therefore, the difference between 100% and the transmittance should be almost equivalent to the absorbance. The variation in the absorption of the second pulse as a function of the delay time is considered to have the same tendency as that of the size of the heat-affected zone and the bonding strength (see Figs. 4 and 6). We conclude that the increases in the size of the heat-affected zone and the bonding strength are due to increased absorption of the second pulse induced by the first pulse. This supports the proposed mechanism based on electron excitation and relaxation processes. To investigate the dependence of the bonding strength on the pulse energy ratio, the delay time employed in the present experiment was 15 ps, which is close to the delay time at which the highest bonding strength was obtained (Fig. 6). It is considered that the first pulse generates free electrons, while the second pulse induces avalanche ionization or electron heating. If the first pulse energy is too low, free electrons will not be efficiently generated and the second pulse will not further excite free electrons. On the other hand, electron heating or avalanche ionization will not occur efficiently if the second pulse energy is too low. Avalanche ionization is particularly affected more by the laser energy. Thus, an energy ratio of unity provides a good balance that permits efficient free electron generation and subsequent excitation, resulting in optimal microwelding. 5. Conclusion We have systematically investigated the dependence of the size of the heat-affected zone and the bonding strength on the delay time for delay time up to 80 ns in microwelding of

#177531 - $15.00 USD

(C) 2012 OSA



photosensitive glass by double-pulse irradiation of an ultrafast laser beam. Both the size of the heat-affected zone and the bonding strength increased rapidly as the delay time was increased from 0 ps to 12.5–15 ps. However, they decreased dramatically when the delay was increased to 30 ps. They had a small peak at around 100 ps and were larger than the singlepulse irradiation for delay time up to 40 ns. The underlying physical mechanism could be explained based on electron excitation and relaxation processes including multiphoton ionization or tunneling ionization, avalanche ionization or electron heating, and absorption by a localized state. Measurements of the transient absorption of the second pulse induced by the first pulse support this consideration. The optimal pulse energy ratio was determined for highefficiency microwelding. These results provide new insights into ultrafast laser microwelding in glass. They are also helpful for understanding the mechanisms of other ultrafast laser processing techniques such as ablation and internal modification of glass.

#177531 - $15.00 USD

(C) 2012 OSA

