September 1, 2003 / Vol. 28, No.

17 / OPTICS LETTERS

1591

Optical breakdown versus filamentation in fused silica by use of femtosecond infrared laser pulses
N. T. Nguyen, A. Saliminia, W. Liu, S. L. Chin, and R. Valle
Centre dOptique, Photonique et Laser, Department of Physics, Universit Laval, Qubec, Canada, G1K 7P4 Received February 17, 2003 The competition between optical breakdown (OB) and laser-pulse filamentation (FL) in bulk fused silica is investigated by using a 1-kHz femtosecond infrared laser. We measure input powers corresponding to the threshold of OB and FL in terms of external focusing conditions. The results demonstrate that OB precedes FL for tight focusing, whereas for sufficiently long focal lengths FL takes places at a lower power than OB does. 2003 Optical Society of America OCIS codes: 320.2250, 260.5950, 140.3440, 160.6030.

The interaction of ultrashort high-power laser pulses with transparent materials has attracted much interest since the advent of powerful femtosecond lasers, particularly owing to the potential for direct writing of three-dimensional optical waveguide structures in bulk glasses.1,2 In fact, a large increase in the refractive index Dn 1022 1023 can be achieved when an intense femtosecond IR laser pulse is focused inside the bulk glass. Although the physical mechanisms underlying such IR photosensitivity are not yet well understood, this technique has been applied to three-dimensional optical storage,3,4 waveguide formation in a wide variety of glasses,5 8 gratings,9 couplers,1,7,10 and photonic crystals.11 Ultrashort laser-pulse-induced optical breakdown (OB) and laser-pulse f ilamentation (FL) are considered to be two electronic excitation processes that can create photostructural modif ications based on plasma formation inside transparent material.7,12,13 However, the extent of such modif ications (e.g., in glass) caused by OB and FL is different. That is, the strong plasma formation associated with OB can lead to permanent damage in the glass structure, whereas the structural modif ications caused by weak plasma formation in the FL process usually give rise to a reversible index change and the formation of good-quality waveguides in glass.13 Filamentation, supercontinuum (SC) generation, and optical breakdown have been extensively discussed in the literature in the past few years.14 18 In particular, the competition between OB and FL in water and in CO2 gas have been reported.19,20 In addition, laser-induced breakdown versus self-focusing in water was numerically studied in the picosecond regime.21 In this Letter we clarify the conditions for the formation of FL and OB and investigate their competition in bulk fused silica in terms of external focusing conditions. The pulses were generated by a Spectra Physics chirped-pulse amplif ication Ti:sapphire laser system at a central wavelength of 810 nm (spectral width of 30 nm) with a 1-kHz repetition rate. The laser beam (diameter 4.6 mm at 1 e2 intensity), after passing through a vacuum spatial filter, was focused by various microscope objectives (Melles-Griot, from 13 to 253) inside a 10-mm-long polished bulk fused silica
0146-9592/03/171591-03$15.00/0

sample. The temporal width of the transform-limited pulses was measured by an optical autocorrelator to be approximately 45 fs before the focusing lens. The input energy was controlled by a half-wave plate placed before the compressor. To determine the thresholds for both OB and FL, we used the experimental setup schematically shown in Fig. 1(a). The appearance of OB with a high-density plasma is usually accompanied by an isotropic broadband plasma emission due to electron recombination, which appears as a white spot around the geometrical focus, as well as scattering and absorption of the input laser beam from the focal region. Initially, a spectrometer (Spectra Pro500i; Acton Research; grating, 300 grooves mm) equipped with an intensified charge-coupled device camera [not shown in Fig. 1(a)] was used to measure the spectrum of the OB signal at different input powers from the side. A lens (BK7, f 60 mm) was used to collect the diverging OB signal, and a f iber bundle was used to couple the collected light onto the entrance slit of the

Fig. 1. (a) Experimental setup. MO, microscope objective; 1, 4, lens; 2, 5, mirror; 3, bandpass filter; 6, blue filter (BG-12); PD, photodiode; PMT, photomultiplier tube. (b) Plasma emission and scattered laser spectra for OB. (c) Typical SC (white-light laser) spectrum. [No filter or mirror used in (b) and (c)]. 2003 Optical Society of America

1592

OPTICS LETTERS / Vol. 28, No. 17 / September 1, 2003

spectrometer. A typical spectrum at 0.5 mJ with a 163 focusing lens f 10.8 mm is shown in Fig. 1(b), where peaks corresponding to plasma emission around 450 nm and laser scattering centered at 810 nm are pronounced. Accordingly, we detected the onset of plasma emission for each objective lens by using a bandpass f ilter with a bandwidth of 10 nm around 450 nm and a photomultiplier tube (Hamamatsu ZC0334) connected to an oscilloscope. Note that, the electron density inside the FL is much lower than that in the OB spot19 and thus was shown to have no inf luence on the measured OB signal from the side. On the other hand, FL occurs because of the balance between the Kerr self-focusing effect (after the instantaneous input power exceeds the critical power 3.77l2 8pn0 n2 , where l, n0 , and n2 are the Pcr wavelength, linear refractive index, and nonlinear refractive index, respectively) and the defocusing of the weak plasma generated through multiphoton excitation (MPE).15 This results in a chirped white-light laser pulse or the so-called supercontinuum, which can be observed by the naked eye in the forward direction.14 The attenuated SC signal was focused onto the entrance slit of the spectrometer. A typical SC spectrum in silica glass that we measured at 5 mJ with a 13 objective lens f 73.5 mm is shown in Fig. 1(c). A blue f ilter (BG-12) with a maximum transmission eff iciency of 60% at 400 nm and a photodiode were used to detect the white-light lasers signal (the so-called FL signal) in the forward direction. The inf luence of OB on the measured FL threshold in the forward direction is neglected, since both 400and 450-nm signals from OB are much lower than the sensitivity level of the photodiode. Two dielectric mirrors with a 0 incident angle at 800 nm were also used to suppress the strong laser line at 810 nm. For both OB and FL we def ined thresholds as the input energy for which the lowest signal was observed in the oscilloscope within the sensitivity level of the detection system used. Enhanced self-phase modulation of an ultrashort pulse in a plasma assisted by the self-steepening phenomenon is commonly regarded as the physical mechanism responsible for SC generation.15,16 Note that self-phase modulation in a neutral medium results in only a symmetric and relatively narrow spectral broadening compared with SC generation, which has a characteristic spectral blueshift extending below 400 nm.15 To properly determine the threshold of FL (or SC generation), we need to quantitatively def ine an appropriate criterion for this process. To date, a qualitative def inition for the FL threshold based on the appearance of a dramatic asymetrical broadening in the spectrum of the light transmitted through the medium has been adapted.20 Our criterion to measure the FL threshold is based on the appearance of a signal around 400 nm. The variation of OB and FL threshold energies in terms of the focal length is shown in Fig. 2. This is our main result. To describe the sharp increase of FL threshold power with decreasing focal length, we present the following physical explanation. It is known that the length of a filament, which can be approximated as the separa-

tion between the self-focus and the geometrical focus, depends on both the pulse peak power and the external focusing condition.15,19 The first self-focusing position starts from the geometrical focus at the critical power and then shifts toward the laser by increasing the input power.15 Thus, with sufficiently short focal lengths, strong OB plasma would be formed at input powers even lower than the critical power for selffocusing, since the required OB threshold intensity can be reached. By further increasing the input power to a level slightly higher than the critical power and for relatively longer focal lengths, the OB generated at the rising part of the pulse can still precede the FL formation. This would happen in the region around the geometrical focus. Subsequent self-focusing of the peak of the pulse would enter the OB plasma zone, which would strongly defocus the pulse; i.e., no FL can be formed inside the OB plasma region around the geometrical focus. However, with increased input energy the self-focal point can be well separated from the region of strong OB plasma. Accordingly, a filament (SC generation) resulting from the balance between the self-focusing and defocusing effect of the weak plasma generated by MPE can be formed outside the geometrical focal volume. As a result of intensity clamping inside the filament,22 the spectral broadening of the white-light laser pulse stops at a wavelength below 400 nm [as seen in Fig. 1(c)]. No further broadening results from further increasing the input energy, while only the spectral density for each blueshifted wavelength in the SC spectrum increases. As shown in Fig. 2, the FL threshold for f 73.5 mm is 0.26 mJ, which is in good agreement with the measured threshold power in the literature.15 Note that no FL signal was observed for focal lengths shorter than 22.5 mm even when the input energy was increased to as high as 8 mJ (to avoid damaging the microscope objective, we did not increase the laser energy to more than 8 mJ). However, by focusing the laser beam in air very close to the front (input) surface of the glass with a 103 objective lens f 16.8 mm , we were able to observe the transmitted SC signal in the forward direction at input energies considerably lower than 8 mJ. In fact, this demonstrates the inf luence of OB in the FL formation. That is, in such

Fig. 2. Threshold energies for OB and FL as a function of focal length. The insets show typical OB and FL plasma images corresponding to each region.

September 1, 2003 / Vol. 28, No. 17 / OPTICS LETTERS

1593

a focusing situation the position of the geometrical focus would be in air; thus no OB plasma would be formed inside the glass. There would be a chance for a laser pulse with a power higher than Pcr to self-focus inside the bulk and to create a plasma due to MPE. As a result, a FL can be formed near the front surface inside the glass, which leads to the observed SC generation. On the other hand, OB is an accumulative process in time that is characterized by a strong plasma with electron densities as high as 1018 1021 cm23 near the focal point.23 At short focal lengths the strong geometrical convergence of the whole beam leads to increased electron density within a very short period of time compared with the duration of the femtosecond pulse because the rate of impact and avalanche ionization processes, which are responsible for OB plasma formation, would be suff iciently high. Since the OB threshold is governed by laser intensity, in a linear propagation regime (no self-focusing) an increasing behavior for the OB threshold power proportional to f 2 is expected. However, as seen in Fig. 2, the OB threshold energy increases more or less linearly with the focal length and even seems to saturate for long focal lengths. This is because self-focusing contributes more significantly in pulse transformation and spot size reduction compared with the case of a tight but linear focusing 7.2 mm we measured geometry. Note that for f an input energy of 0.09 mJ I 1014 W cm22 for the OB threshold, which is in good agreement with the OB threshold intensity for silica reported in the literature.23 To present an applied aspect of the results, we def ine four regions as illustrated in Fig. 2. In region 1 no OB or FL is formed; in other words, no photoinduced modifications can be created in the bulk glass. Region 2 corresponds to the energy and focal lengths for which only OB occurs and no FL is formed. As a result, one can write buried waveguides in bulk fused silica based on the pure OB process. FL occurs at a lower power compared with OB in region 4. This can allow fabrication of smooth waveguides with improved guiding characteristics based on the FL process. Obviously, in this region the required power for OB formation is higher than the critical power for self-focusing. The coexistence of FL and OB takes place for input energies and focal lengths in region 3. Note that in this region, since both OB and FL are present, the spectrum of OB can be partially modif ied by the inf luence of the FL spectrum and vice versa. However, since the OB spectral density is considerably lower than that of SC, its inf luence on the FL spectrum is not significant. The typical images of FL and OB plasma obtained with the CCD camera from the side for the three illustrated regions are also shown in the insets in Fig. 2. In summary, we have investigated the formation conditions of OB and FL in bulk fused silica with focused femtosecond IR pulses. The OB and FL threshold energies were measured in terms of the external focusing conditions, and the observed dependencies were qualitatively described. These experimental

results enable us to properly separate writing regimes corresponding to FL and OB processes to photoimprint efficient three-dimensional waveguide structures in bulk silica glasses. This research was partially supported by the consortium FemtoTech, the Canadian Institute for Photonic Innovations, Fonds pour la Formation de Chercheurs et lAide la Recherche, the Natural Sciences and Engineering Research Council, the Canada Foundation for Innovation, Defence Research and Development Canada Valcartier, and the Canada Research Chair. A. Saliminias e-mail address is asalimin@phy.ulaval.ca. References
1. K. Minoshima, A. M. Kowalevicz, I. Hartl, E. Ippen, and J. G. Fujimoto, Opt. Lett. 26, 1516 (2001). 2. A. Marcinkevicius, S. Juodkazis, M. Watanabe, M. Miwa, S. Matsuo, H. Misawa, and J. Nishii, Opt. Lett. 26, 277 (2001). 3. E. N. Glezer, M. Milosavljevic, L. Huang, R. J. Finlay, T. H. Her, J. P. Callan, and E. Mazur, Opt. Lett. 21, 2023 (1996). 4. W. Watanabe, T. Toma, K. Yamada, J. Nishii, K.-I. Hayashi, and K. Itoh, Opt. Lett. 25, 1669 (2000). 5. K. M. Davis, K. Miura, K. Sugimoto, and K. Hirao, Opt. Lett. 21, 1729 (1996). 6. A. Saliminia, N. T. Nguyen, M. C. Nadeau, S. Petit, S. L. Chin, and R. Valle, J. Appl. Phys. 93, 3724 (2003). 7. D. Homoelle, S. Wielandy, A. L. Gaeta, N. F. Borrelli, and C. Smith, Opt. Lett. 24, 1311 (1999). 8. Y. Sikorski, A. A. Said, P. Bado, R. Maynard, C. Florea, and K. A. Winick, Electron. Lett. 36, 226 (2000). 9. E. Fertein, C. Przygodzki, H. Delbarre, A. Hidayat, M. Douay, and P. Niay, Appl. Opt. 40, 3506 (2001). 10. A. M. Streltsov and N. F. Borrelli, Opt. Lett. 26, 42 (2001). 11. Y. Xu, S. Juodkazis, K. Sun, M. Wabe, S. Matsuo, H. Misawa, and J. Nishii, Opt. Lett. 26, 325 (2001). 12. S. H. Cho, H. Kumagai, and K. Midorikawa, Opt. Commun. 207, 243 (2002). 13. K. Yamada, W. Watanabe, T. Toma, K. Itoh, and J. Nishii, Opt. Lett. 26, 19 (2001). 14. S. L. Chin, A. Brodeur, S. Petit, O. G. Kosareva, and V. P. Kandidov, J. Nonlinear Opt. Phys. Mater. 8, 121 (1999). 15. A. Brodeur and S. L. Chin, J. Opt. Soc. Am. B 16, 637 (1999). 16. N. Akozbek, M. Scalora, C. M. Bowden, and S. L. Chin, Opt. Commun. 191, 353 (2001). 17. P. K. Kennedy, IEEE J. Quantum Electron. 31, 2250 (1995). 18. D. Du, X. Liu, G. Korn, J. Squier, and G. Mourou, Appl. Phys. Lett. 64, 3071 (1994). 19. W. Liu, O. Kosareva, I. S. Golubtsov, A. Iwasaki, A. Becker, V. P. Kandidov, and S. L. Chin, Appl. Phys. B 76, 215 (2003). 20. F. A. Ilkov, L. S. Ilkova, and S. L. Chin, Opt. Lett. 18, 681 (1993). 21. Q. Feng, J. V. Moloney, A. C. Newell, and E. M. Wright, Opt. Lett. 20, 1958 (1995). 22. W. Liu, S. Petit, A. Becker, N. Akozbek, C. M. Bowden, and S. L. Chin, Opt. Commun. 202, 189 (2002). 23. M. Li, S. Menon, J. P. Nibarger, and G. N. Gibson, Phys. Rev. Lett. 11, 2394 (1999).