Absorption mechanism of the second pulse in ... - OSA Publishing

3 downloads 2365 Views 1MB Size Report
are 104.2 and 714.3 ps, respectively. Single-photon absorption of the second pulse by free electrons dominates in domain I, resulting in high bonding strength.
Absorption mechanism of the second pulse in double-pulse femtosecond laser glass microwelding Sizhu Wu,1,2 Dong Wu,1 Jian Xu,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: The absorption mechanism of the second pulse is experimentally and theoretically investigated for high-efficiency microwelding of photosensitive glass by double-pulse irradiation using a femtosecond laser. The transient absorption change during the second pulse irradiation for various energies induced by the first pulse is measured at different delay times. The resulting effects depend on whether the delay time is 0–30 ps (time domain I) or 30– several ns (domain II). By solving rate equations for the proposed electronic processes, the excitation and relaxation times of free electrons in time domain I are estimated to be 0.98 and 20.4 ps, respectively, whereas the relaxation times from the conduction band to a localized state and from the localized state to the valence band in domain II are 104.2 and 714.3 ps, respectively. Single-photon absorption of the second pulse by free electrons dominates in domain I, resulting in high bonding strength. In time domain II, about 46% of the second pulse is absorbed by a single photon due to the localized state, which is responsible for higher bonding strength compared with that prepared by single-pulse irradiation. ©2013 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. 8. 9.

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). 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. 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, 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). 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). K. Cvecek, I. Miyamoto, J. Strauss, V. Bui, S. Scharfenberg, T. Frick, and M. Schmidt, “Strength of joining seam in glass welded by ultra-fast lasers depending on focus height,” J. Laser Micro/Nanoeng 7(1), 68–72 (2012).

#193735 - $15.00 USD Received 11 Jul 2013; revised 9 Sep 2013; accepted 10 Sep 2013; published 1 Oct 2013 (C) 2013 OSA 7 October 2013 | Vol. 21, No. 20 | DOI:10.1364/OE.21.024049 | OPTICS EXPRESS 24049

10. 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). 11. 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). 12. 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). 13. D. Helie, F. Lacroix, and R. Vallee, “Reinforcing a direct bond between optical materials by filamentation based femtosecond laser welding,” J. Laser Micro/Nanoeng. 7(3), 284–292 (2012). 14. 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). 15. 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). 16. 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). 17. S. Wu, D. Wu, J. Xu, Y. Hanada, R. Suganuma, H. Wang, T. Makimura, K. Sugioka, and K. Midorikawa, “Characterization and mechanism of glass microwelding by double-pulse ultrafast laser irradiation,” Opt. Express 20(27), 28893–28905 (2012). 18. 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). 19. 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). 20. M. Shimizu, M. Sakakura, M. Ohnishi, M. Yamaji, Y. Shimotsuma, K. Hirao, and K. Miura, “Three-dimensional temperature distribution and modification mechanism in glass during ultrafast laser irradiation at high repetition rates,” Opt. Express 20(2), 934–940 (2012). 21. I. Alexeev, K. Cvecek, C. Schmidt, I. Miyamoto, T. Frick, and M. Schmidt, “Characterization of shear strength and bonding energy of laser produced welding seams in glass,” J. Laser Micro/Nanoeng. 7(3), 279–283 (2012). 22. K. Sugioka, S. Wada, H. Tashiro, K. Toyoda, Y. Ohnuma, and A. Nakamura, “Multiwavelength excitation by vacuum-ultraviolet beams coupled with fourth harmonics of a Q-switched Nd:YAG laser for high-quality ablation of fused quartz,” Appl. Phys. Lett. 67(19), 2789–2791 (1995). 23. K. Nahen and A. Vogel, “Plasma formation in water by picosecond and nanosecond Nd:YAG laser pulses-part II: transmission, scattering, and reflection,” IEEE J. Sel. Top. Quantu. Electron. 2(4), 861–871 (1996). 24. 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). 25. 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. Phys. Res. B 116, 43–48 (1996). 26. K. S. Song and R. T. Williams, Self-Trapped Excitons (Springer-Verlag, Berlin, Heiderberg, 1993). 27. B. Fisette and M. Meunier, “Three-dimensional microfabrication inside photosensitive glasses by femtosecond,” J. Laser Micro/Nanoeng. 1(1), 7–11 (2006). 28. J. Kim, H. Berberoglu, and X. Xu, “Fabrication of microstructures in photoetchable glass ceramics using excimer and femtosecond lasers,” J. Micro. Nanolith. 3(3), 478–485 (2004).

1. Introduction Glass microwelding induced by an ultrafast laser has been drawing attention for applications such as optics, microelectromechanical systems, precision machinery, healthcare, and small satellites. Many groups have successfully demonstrated microwelding of various glass including borosilicate [1–9], fused silica [10–14], photosensitive glass [15–17], and others [18–21]. In order to enhance the efficiency of glass microwelding, a new strategy [16] has been proposed using double-pulse irradiation by an ultrafast laser beam. This technique increases the bonding strength of photosensitive glass welding by 22% relative to that for conventional single-pulse irradiation. The physical mechanism of glass microwelding by double-pulse irradiation has been systematically investigated by measuring the dependence of the size of the heat-affected zone and the bonding strength on the delay time between the two pulses and analyzing the transient absorption induced by ultrafast laser pulses using a pumpprobe technique [17]. An increase in both the heat-affected zone and the bonding strength was found to be associated with an increase in the second-pulse absorption induced by the first pulse. The first pulse induces multiphoton or tunneling ionization, while the second pulse induces electron heating or avalanche ionization with a delay of up to 30 ps. Alternatively, the second pulse is absorbed by the localized state with a delay of as much as several

#193735 - $15.00 USD Received 11 Jul 2013; revised 9 Sep 2013; accepted 10 Sep 2013; published 1 Oct 2013 (C) 2013 OSA 7 October 2013 | Vol. 21, No. 20 | DOI:10.1364/OE.21.024049 | OPTICS EXPRESS 24050

nanoseconds. However, there remain issues to be clarified in order to optimize the process, such as the excitation and relaxation times of the excited electrons and how many photons are necessary for the second pulse absorption. In this paper, the mechanism is systematically investigated by measuring the absorption of the second pulse for different laser powers as a function of the delay times after the first pulse irradiation. The second pulse absorption is different during the first 30 ps than it is during the next few nanoseconds. Therefore, the transient absorption change of the second pulse is decomposed into two time domains. The electronic excitation time from the valence to the conduction band, and the relaxation time of free electrons in the conduction band and of electrons trapped in the localized state are estimated by fitting the experimental results using rate equations. 2. Experimental setup for evaluation of the second-pulse absorption

Fig. 1. Experimental setup to measure the transient absorption of the second pulse at different delay times.

The substrates are commercially available photosensitive glass (Foturan from Schott Glass) that consists of lithium aluminosilicate doped with trace amounts of silver, cerium, sodium, and antimony. An amplified femtosecond Er-fiber laser system (IMRA model FCPA μJewel D-400) generates 360-fs pulses at a wavelength of 1045 nm with a repetition rate of 200 kHz. The experimental setup is the same as that used for previous glass microwelding studies [16,17] except for the setup to measure the power of the transmitted laser pulses. The delay time between the two pulses is controlled by adjusting the optical path in the delay line using a high-precision stage, as confirmed by cross-correlation. The first pulse power is fixed at 155 mW, the power used for glass welding, while the second pulse power is varied from 5 to 155 mW. The p-polarized first and the s-polarized second pulses are focused into the glass substrate using an objective lens having a numerical aperture (NA) of 0.4. Figure 1 illustrates the scheme for measuring the energy of the second pulse transmitted by the photosensitive glass at different delay times. A polarized beam-splitter placed behind the glass substrate reflects the s-polarized second pulse to a power meter. 3. Estimate of the electronic excitation and relaxation times It is considered that deposition of as much laser energy as possible into the glass results in efficient heating to produce a large molten zone for high-quality welding. Therefore, exploration of the electronic excitation and relaxation times is important for efficient transfer of the energy of the second pulse to the glass.

#193735 - $15.00 USD Received 11 Jul 2013; revised 9 Sep 2013; accepted 10 Sep 2013; published 1 Oct 2013 (C) 2013 OSA 7 October 2013 | Vol. 21, No. 20 | DOI:10.1364/OE.21.024049 | OPTICS EXPRESS 24051

Optical density

0.6

Experimental data

0.5 0.4

I

II

0.3

0.00

0

100

200 300 Delay time (ps)

400

Fig. 2. Dependence of the optical density of the second pulse on the delay time. It is decomposed into two time domains: 0–30 ps (domain I) and 30–400 ps (domain II).

Using rate equations, based on previously proposed electronic processes in the photosensitive glass [22], the excitation and relaxation times are estimated by fitting the experimental data. The second pulse absorption differs in two delay time domains. Prior to 30 ps, the second pulse is absorbed by free electrons generated by the first pulse, resulting in electron heating or avalanche ionization, which imparts a high bonding strength during glass welding. Between 30 ps and a couple of ns, during which the bonding strength remains higher than that prepared by conventional single-pulse irradiation, the second pulse is absorbed by electrons trapped at a localized state. Figure 2 shows that a plot of optical density (OD) of the second pulse as a function of the delay time can be divided into these two time domains. The optical density is calculated from the measured transmittance of the second pulse OD = – log10(IT/I0), where I0 and IT are the incident and transmitted laser energies, respectively. The incident powers of the first and second pulses are 155 mW at 200 kHz. An assumption is made that the reflection and scattering by the laser-induced plasma are negligible based on the results of Nahen et al. [23]. Our previous experiments confirm that the reflection of the second pulse is nearly independent of the delay time. Therefore, the transmittance and absorbance add up to about 100%.

#193735 - $15.00 USD Received 11 Jul 2013; revised 9 Sep 2013; accepted 10 Sep 2013; published 1 Oct 2013 (C) 2013 OSA 7 October 2013 | Vol. 21, No. 20 | DOI:10.1364/OE.21.024049 | OPTICS EXPRESS 24052

Fig. 3. (a) Two-level model for time domain I associated with free-electron generation and relaxation. (b) Three-level model for domain II associated with electron trapping at a localized state such as a defect or exciton.

During time domain I, the two-level model of Fig. 3(a) is adopted, because free-electron generation and relaxation in the conduction band dominate. The electrons are excited from the valence to the conduction band by the first pulse. Although the interband excitation in photosensitive glass is elaborate as discussed later, a simple process, i.e., direct excitation from the valence band to the conduction band, is considered here to calculate the time for free electron generation. The resulting free electrons interact with the second pulse to weld the glass with a high bonding strength. The electrons eventually relax out of the conduction band. Therefore, it’s very important to explore the relaxation time of free electrons to determine the optimum timing of the second pulse. The rate equation of the two-level model is in general given by [22]

dN c ( t ) / dt =  ΓI n ( t ) + γ vc  N v ( t ) − γ cv N c ( t )

(1)

where Γ is the electronic excitation rate induced by the femtosecond pulses of intensity I ( t ) and n is the number of photons necessary for inducing absorption. In the photosensitive glass, the interband excitation occurs by three-photon absorption as discussed later, which gives n = 3 in Eq. (1). Given that the full width at half maximum (FWHM) of the laser pulses is 360 fs, 2 I ( t ) = exp  − ( t − FWHM ) / 2σ 2  σ 2π where FWHM ≈ 2.35σ . Here γ vc is the thermal   excitation rate, γ cv is the relaxation rate, and N v ( t ) and N c ( t ) are the electron densities at time t in the valence and conduction band, respectively, where N c ( t ) + N v ( t ) = 1 with N v ( 0 ) = 1 and N c ( 0 ) = 0 . Owing to the large energy gap of the photosensitive glass, γ vc can

be ignored in the current case. In this time domain, free-electron absorption by the second pulse is dominant. Thus, the time-dependent variation of the OD of the second pulse is proportional to the free-electron density in the conduction band N c ( t ) . According to the

3 least-squares fitting in time domain I , the excitation time 1 ( ΓI max ) and the relaxation time

1 γ cv are therefore 0.98 and 20.4 ps, respectively. Accordingly, to achieve high-quality microwelding, the second pulse should arrive at a delay time between 1 and 20 ps. During this time interval, the number of electrons in the conduction band is maximal and most of the second laser pulse energy will be absorbed. In agreement with this conclusion, the highest bonding strength was achieved for delay times of 1–15 ps in our previous experiment. The excitation time is longer than the few hundred femtoseconds observed for typical glasses,

#193735 - $15.00 USD Received 11 Jul 2013; revised 9 Sep 2013; accepted 10 Sep 2013; published 1 Oct 2013 (C) 2013 OSA 7 October 2013 | Vol. 21, No. 20 | DOI:10.1364/OE.21.024049 | OPTICS EXPRESS 24053

because of the complexity of the interband excitation of electrons through various defect levels by multiphoton absorption [24]. In addition, the relaxation time is much longer than 0.5 ps for fused silica [25], perhaps because the photosensitive glass has a more complex system than does fused silica. Next, we simulate the relaxation times in the time domain II where the bonding strength is still higher than that for the conventional single-pulse irradiation sample. The absorption of the second pulse in this time domain arises from electrons trapped at a localized state such as a defect or exciton. Therefore, the three-level model of Fig. 3(b) is adopted in this domain. For simplicity, set N c ( 0 ) = N , N  ( 0 ) = 0, and N v ( 0 ) = 0 because the electronic excitation takes place much earlier than the start of this time domain. The rate equations then become dN c ( t ) = − rc N c ( t ) , dt dN  ( t ) = + rc N c ( t ) − rv N  ( t ) , and dt dN v ( t ) = + rv N  ( t ) dt

(2)

where N  ( t ) is the electron density at time t in the localized state, γ c is the relaxation rate from the conduction band to the localized state, and γ v is the relaxation rate from the localized state to the valence band. By solving Eq. (2), the electron density in the localized state is found to be N (t ) =

N exp ( − rv t ) − exp ( − rc t )  1 − rv rc 

(3)

The variation of the OD in this time domain is proportional to N  ( t ) because the absorption of the second pulse is due to electrons trapped at the localized state. Therefore, the relaxation time from the conduction band to the localized state and from the localized state to the valence band can be obtained by fitting Eq. (3) to Fig. 2, resulting in 104.2 and 714.3 ps, respectively. This 104.2 ps relaxation time from the conduction band to the localized state is much longer than the 20.4 ps from the conduction band to the valence band calculated above. However, the relaxation rate of free electrons out of the conduction band is expected to be constant. Thus, there may be another intermediate state that has a relaxation time of a few tens of picoseconds. The electrons could first relax from the conduction band to the intermediate state with a time of ~20 ps, and then further decay to the localized state with a relaxation time of ~80 ps. This idea would explain the small peak observed near 100 ps in Fig. 2. The relaxation time from the localized state to the valence band is calculated to be 714.3 ps. One of the possible origins for the localized state may therefore be a self-trapped exciton, because the excitons have a typical lifetime of about 1 ns in silica glass at room temperature [26]. Another possibility is the lattice defect.

#193735 - $15.00 USD Received 11 Jul 2013; revised 9 Sep 2013; accepted 10 Sep 2013; published 1 Oct 2013 (C) 2013 OSA 7 October 2013 | Vol. 21, No. 20 | DOI:10.1364/OE.21.024049 | OPTICS EXPRESS 24054

4. Determination of the number of photons for the second pulse absorption

100

prob 155 mW prob 125 mW prob 100 mW prob 75 mW

Absorption (%)

80

prob 50 mW prob 25 mW prob 15 mW prob 5 mW

60 40 20 0

0

50 100 150 200 250 300 350 400 Delay time (ps)

Fig. 4. Dependence of the transient absorption of the second pulse on the delay time for different second pulse powers ranging from 5 to 155 mW.

Multiphoton absorption inducing the interband excitation is dominant for conventional singlepulse irradiation. However, double-pulse irradiation can more efficiently deposit the laser energy in the glass, resulting in higher bonding strength. To demonstrate the effect, the number of photons absorbed during the second pulse irradiation is estimated in each time domain. For this estimation, the dependence of the transient absorption during the second pulse irradiation on the delay time for different second pulse powers was measured as shown in Fig. 4. 4.1 The number of photons absorbed in time domain I

Fig. 5. Absorption channels for the second pulse in time domain I. One channel is interband excitation via the defect level, and the other channel is absorption by free electrons.

#193735 - $15.00 USD Received 11 Jul 2013; revised 9 Sep 2013; accepted 10 Sep 2013; published 1 Oct 2013 (C) 2013 OSA 7 October 2013 | Vol. 21, No. 20 | DOI:10.1364/OE.21.024049 | OPTICS EXPRESS 24055

2 4

Absorbed power density (10 w/cm

(a)

)

The photosensitive glass has an elaborate electronic excitation process due to its complex composition. The electrons in the valence band are first excited to an intermediate state and then they are further excited to the conduction band to generate free electrons [24], resulting in the two-step excitation process depicted in Fig. 5. When the second pulse is incident at a delay time of between 1 and 20 ps (which are the excitation and relaxation times of free electrons calculated above), there are two possible absorption channels for the second pulse. One possibility is interband excitation via a defect level, and the other channel is absorption by free electrons. For the first step of the interband process [24, 27], an energy larger than 3.3 (3.9) eV is necessary, corresponding to 3 (4) photons of the femtosecond laser. For the second step, between 3.2 and 4.8 eV is needed, corresponding to between 3 and 5 photons. Meanwhile, free-electron absorption of a single photon occurs. Therefore, single-, three-, four- and five-photon absorption processes are possible for the second pulse in time domain I.

70 Experiment Simulation

60 50 40 30 20 10 0

0

20 40 60 80 100 120 140 160 nd

2 4

Absorbed power density (10 w/cm

(b)

)

Incident power of 2 pulse (mW)

35

Single photon absorption Three photon absorption Four photon absorption Five photon absorption

30 25 20 15 10 5 0

0

20 40 60 80 100 120 140 160 nd

Incident power of 2 pulse (mW) Fig. 6. (a) Dependence of the absorbed power density of the second laser pulse on the incident laser power at a delay time of 15 ps obtained by experimental measurement (black line) and by data fitting (red line). (b) Laser power density absorbed by each absorption process for different second pulse powers.

Based on these considerations, absorption of the second pulse is given by [28]

#193735 - $15.00 USD Received 11 Jul 2013; revised 9 Sep 2013; accepted 10 Sep 2013; published 1 Oct 2013 (C) 2013 OSA 7 October 2013 | Vol. 21, No. 20 | DOI:10.1364/OE.21.024049 | OPTICS EXPRESS 24056

dI = −α1 I − α 3 I 3 − α 4 I 4 − α 5 I 5 dx

(4)

where I is the laser intensity in W/cm2 at x, α n is the nonlinear absorption coefficient for nphoton absorption, and x is the distance from the surface in the substrate. Figure 6(a) plots the dependence of the absorbed power density of the second pulse on the incident laser power at a delay time of 15 ps where the largest bonding strength occurs. The black curve is derived from Fig. 4, whereas the red line is computed by fitting the data to Eq. (4). For the fitting, dI/dx was calculated as (IT – I0)/DOF where I0, IT are the incident and the transmitted laser intensities and DOF (Depth of Focus) = λ/NA2, and I was set at the laser intensity around the center of focal volume (I = (I0 + IT)/2). The fit is excellent and the absorption coefficients for single-, three-, four-, and five-photon absorption are calculated to be 9.96 × 102 cm–1, 3.19 × 10−9 cm3 W–2, 2.88 × 10−16 cm5 W–3, and 1.29 × 10−22 cm7 W–4, respectively. Using these absorption coefficients, Fig. 6(b) graphs the laser power density absorbed in each process as a function of the incident power of the second pulse. This result indicates that single- and three-photon absorption are much more likely than four- or five-photon absorption. Thus, one concludes that interband excitation takes place by three-photon absorption in both steps, whereas free-electron absorption only by single photons. In the current case, the first pulse energy is only used for free electron generation since the first pulse cannot interact with free electrons due to the significantly longer excitation time of 0.98 ps (see Sec. 3) than the pulse width of 360 fs. In the meanwhile, once the free electrons are generated by the first pulse, the second pulse is more probably absorbed by the free electrons than interband excitation through the intermediate state due to single photon absorption. In fact, about 55% of the second laser energy is absorbed by the free electrons even at the largest power of 155 mW used for glass welding. This result explains why the highest bonding strength is obtained at a delay time of around 15 ps. 4.2 The number of photons absorbed in time domain II

EC 3 hv 1 hv Intermediate state

3 hv

Excited localized state Localized state

EV Fig. 7. Absorption channels for the second pulse in time domain II. The second pulse can be absorbed by the interband excitation and by the electrons trapped at the localized state.

Between 30 ps and a couple of ns, absorption of the second pulse by electrons trapped at the localized state continues to result in higher bonding strength compared with that prepared by conventional single-pulse irradiation [17]. In this time domain, the second pulse can be absorbed both by the interband excitation and by the electrons trapped at the localized state, as shown in Fig. 7. The interband excitation is induced by the three-photon absorption discussed in Sec. 4.1. To achieve a higher bonding strength, the localized state absorption

#193735 - $15.00 USD Received 11 Jul 2013; revised 9 Sep 2013; accepted 10 Sep 2013; published 1 Oct 2013 (C) 2013 OSA 7 October 2013 | Vol. 21, No. 20 | DOI:10.1364/OE.21.024049 | OPTICS EXPRESS 24057

needs to be more efficiently induced than the interband excitation. Therefore, the number of photons required for the localized state absorption should be smaller than 3. Based on this consideration, the data are fit to the following four combinations at a delay time of 100 ps corresponding to the small peak in Fig. 2, 1−, 2−, 3 − photon absorption :

dI = −α1 I − α 2 I 2 − α 3 I 3 dx

(5)

dI = −α1 I − α 3 I 3 dx

(6)

1−, 3 − photon absorption : 2−, 3 − photon absorption :

dI = −α 2 I 2 − α 3 I 3 , and dx dI = −α 3 I 3 . dx

Experiment Simulation 1,2,3 photon

50 40

2

50 40

30

30

20

20

10

10 0

20 40 60 80 100 120 140 160 0

20 40 60 80 100 120 140 160 Incident power of 2 pulse (mW)

Incident power of 2 pulse (mW)

70 60

)

70

0

nd

2

0

4

) 2 4

60

Experiment Simulation 1,3 photon

4

60

(b)

Absorbed power density (0 w/cm

(a)

nd

Absorbed power density (10 w/cm

70

)

70

(8)

(c)

(d)

Experiment Simulation 2, 3 photon

60 50

Experiment Simulation 3 photon

40

Absorbed power density (10 w/cm

4

Absorbed power density (10 w/cm

2

)

3 − photon absorption :

(7)

50 40

30

30

20

20

10

10

0

0

20 40 60 80 100 120 140 160 0 nd

Incident power of 2 pulse (mW)

20 40 60 80 100 120 140 160

0

nd

Incident power of 2 pulse (mW)

Fig. 8. Dependence of the absorbed power intensity of the second pulse on the incident laser power. Black lines: experimental results; red lines: data fitting using (a) Eq. (5) for single-, two-, and three-photon absorption, (b) Eq. (6) for single- and three-photon absorption, (c) Eq. (7) for two- and three-photon absorption, and (d) Eq. (8) for only three-photon absorption.

The results are plotted in Fig. 8 as the red lines. The combination of 1- and 3-photon absorption in Fig. 8(b) gives the best fit to the experimental results from Fig. 4 graphed by the black line. These results indicate that localized state absorption occurs via a single photon, since the interband excitation is induced by three photons. Consequently, the localized state is more efficiently excited than are free electrons by three-photon absorption. This explains why a higher bonding strength is observed in time domain II than for conventional single-pulse irradiation.

#193735 - $15.00 USD Received 11 Jul 2013; revised 9 Sep 2013; accepted 10 Sep 2013; published 1 Oct 2013 (C) 2013 OSA 7 October 2013 | Vol. 21, No. 20 | DOI:10.1364/OE.21.024049 | OPTICS EXPRESS 24058

) 2 4

Absorbed power density (10 w/cm

35 Single photon absorption Three photon absorption

30 25 20 15 10 5 0

0

20 40 60 80 100 120 140 160 nd

Incident power of 2 pulse (mW) Fig. 9. Laser power density absorbed by single- and three-photon processes for different second pulse powers estimated using the calculated absorption coefficients at a delay time of 100 ps.

The absorption coefficients calculated using Eq. (6) are 8.29 × 102 cm–1 for single-photon absorption and 3.37 × 10−9 cm3 W–2 for three-photon absorption. Figure 9 plots the laser power density absorbed by single and three photons for different second pulse powers using the calculated absorption coefficients at a delay time of 100 ps. Even at the largest power of 155 mW, as used for glass welding, about 46% of the energy of the second pulse is absorbed by the localized state to form the excited-localized state. There are three channels for relaxation of the excited-localized state, which are nonradiative transitions to its grand state and the valence band and a radiative transition to its grand states. The former two processes efficiently transform the absorbed laser energy into heat, while the latter one is nonthermal process. Since the radiative process usually occurs much later than the nonradiative processes, the nonradiative process is much more efficiently induced [26], resulting in more efficient heat generation which is responsible for a higher bonding strength than for single-pulse irradiation. 5. Conclusion

The physical mechanism of photosensitive glass microwelding by femtosecond double-pulse irradiation was experimentally and theoretically investigated by measuring the transient absorption of the second pulse induced by the first pulse. The second-pulse absorption arises from electronic excitation and relaxation processes that depend on the delay time in two different domains. In time domain I, ranging from 0 to 30 ps, the excitation and relaxation times of the free electrons were estimated to be 0.98 and 20.4 ps, respectively, whereas in time domain II (from 30 ps to several ns), the relaxation time from the conduction band to the localized state and from the localized state to the valence band were calculated to be 104.2 and 714.3 ps, respectively. Furthermore, we found that single-photon absorption of the second pulse by free electrons dominates in time domain I, which results in the highest possible bonding strength. Meanwhile in time domain II, the second pulse is absorbed by the localized state by a single photon, and approximately 46% of its energy is transferred to the localized state at the power of 155 mW that is used for glass welding. This localized state absorption results in a higher bonding strength than for conventional single-pulse irradiation. The singlephoton absorption of the 1045-nm beam by the localized state occurs during time domain II relative to the arrival of the first pulse. These results suggest a new method of combining femtosecond irradiation with a nanosecond laser for efficient microwelding of glass. #193735 - $15.00 USD Received 11 Jul 2013; revised 9 Sep 2013; accepted 10 Sep 2013; published 1 Oct 2013 (C) 2013 OSA 7 October 2013 | Vol. 21, No. 20 | DOI:10.1364/OE.21.024049 | OPTICS EXPRESS 24059