Appl Phys A (2011) 102: 49–54 DOI 10.1007/s00339-010-6083-4
R A P I D C O M M U N I C AT I O N
Shadowgraphic imaging of laser transfer driven by metal film blistering T.V. Kononenko · P. Alloncle · V.I. Konov · M. Sentis
Received: 19 April 2010 / Accepted: 7 October 2010 / Published online: 26 October 2010 © Springer-Verlag 2010
Abstract A time-resolved shadowgraphic technique was used to investigate local transfer of diamond nanopowder from a thin titanium film on a silica support under irradiation by 450-fs or 50-ps laser pulses. Clean powder ejection driven by blistering of the metal film remaining on the target surface was found possible in a limited fluence range, but the metal film was removed from the target together with the powder when higher laser fluences were applied. The velocity of the powder ejection demonstrates an approximately linear rise in a wide range of the incident fluences, while the slope of the velocity curve decreases for thicker metal layers. The maximum ejection velocity achievable in the blistering regime was evaluated as ∼100 m/s independently of the metal thickness and pulse width.
1 Introduction Laser-induced forward transfer (LIFT) was proposed initially to transport small pieces of a thin metal film on a receiving substrate placed in close proximity to the irradiated target [1]. Later, a metal film deposited on a transparent support was found useful for LIFT of various fragile substances like polymers, proteins, microorganisms, etc. [2–6]. A significant function of the metal layer in this technique is efficient absorption of the laser radiation in a wide wavelength range. Besides, the metal film participates in local ejection T.V. Kononenko () · V.I. Konov General Physics Institute, Vavilov str. 38, 119991 Moscow, Russia e-mail:
[email protected] Fax: +7-499-5038151 P. Alloncle · M. Sentis LP3 Université Aix-Marseille II, Luminy, Case 917, 13288 Marseille Cedex 9, France
of the main material via one of two mechanisms, as a rule. First, transfer of a solution or a suspension can be initiated by boiling of liquid within a thin layer bordering the laserheated metal [2, 3]. Second, both solid and liquid materials can be transferred due to laser-induced ablation of the metal film [4–6]. To minimize unavoidable contamination of the receiving substrate by metal microdroplets in the latter case, a very thin metal film is preferably used. We have reported recently a so-called blister-based laserinduced forward transfer (BB-LIFT) technique utilizing a new principle. A dry diamond nanopowder [7] and organic molecules [8] were locally transferred from targets covered by a thin metal film, avoiding total laser sputtering of the metal. Material ejection results from fast blistering of the metal film, which remains on the target. Main advantages of the BB-LIFT procedure compared with other LIFT techniques are applicability for liquid-free substances, negligible contamination of the transferred material by ablation products and opportunity to minimize heating of the transferred material down to zero. Laser-induced deformation of a metal film was utilized earlier in experiments on delivering drug-coated metal microparticles into soft body targets [9, 10]. Intensive laser ablation of the front side of a 100-µm-thick, free-standing aluminum foil generated a powerful shock wave propagating through the foil and accelerating microparticles on its back side. The crucial distinction of the BB-LIFT technique from the mentioned approach is usage of much thinner metal films (0.05–2 µm) requiring a special support. Such modification dramatically improves spatial resolution of the transfer procedure, making it competitive with other LIFT techniques. Let us note also the resemblance of the considered BB-LIFT technique to the laser transfer process driven by plastic deformation of a 4-µm-thick polyimide film [11].
50
Time-resolved shadow imaging was found very useful in the study of ablative laser transfer of thin metal films [12–14] and microparticles [10], oxides [15], viscous inks [16, 17] and different organic materials [5, 18]. It has been shown that the produced ablative plume propagates outward of the target at a speed which increases with incident laser fluence until laser damage of the transparent substrate stops this trend [12]. The maximum reported velocity of the plume front reached 2 km/s [12, 13]. Applying the laser-induced fluorescence method, Nakata and Okada [13] have found that such high velocities relate only to excited atoms, while large emissive particles in the plume propagate much slower (∼100 m/s). However, later experiments have revealed that laser ablation allows acceleration of micrometer-scale objects to much higher speeds. In particular, Hopp et al. [5] observed ablative ejection of conidia from a metalized target surface at a speed of 1150 m/s. According to Young et al. [17], laser ablation of viscous inks can produce a cloud of droplets accelerated up to 600 m/s. Generation of a powerful shock wave in a free-standing aluminum foil ablated by high-energy (1.4 J) nanosecond pulses results in ejection of metal microparticles with a speed estimated by Menezes et al. [10] at the level of 1.6 km/s. Here we first report time-resolved shadowgraphic imaging of the transfer process driven by blistering of a thin metal film. Liquid-free ejection of diamond nanopowder from targets covered by titanium films was initiated by short laser pulses of two pulse widths—450 fs and 50 ps. The effect of the laser fluence and the metal film thickness on the ejection velocity was investigated. The physical mechanisms providing the metal film blistering at different conditions are discussed.
2 Experiments A few targets were prepared by depositing titanium films of different thicknesses (50–1700 nm) on polished silica plates via vacuum evaporation. Ultra-dispersed diamond (UDD) powder obtained by the explosive method [19] was suspended in distilled water with dissolved surfactant (sodium dodecyl sulphate, 2.0 mg/ml). The suspension was subjected to ultrasonic processing and following sedimentation to remove large particles. A small portion of the suspension was spread over the metalized surface of the prepared targets. After evaporation of water, we obtained a continuous, ∼1-µmthick layer of the diamond particles with a maximum size of ∼200 nm according to atomic force microscopy measurements. Two short-pulse laser systems were involved in the experiments: a Nd:YAG laser (‘Continuum’) generating 50-ps pulses at λ = 1064 nm and a femtosecond laser (Amplitude System) providing 450-fs pulses at λ = 1025 nm. The experiments were performed in the standard air environment.
T.V. Kononenko et al.
Fig. 1 Setup for time-resolved shadowgraphic imaging with picosecond laser
The pulsed laser radiation was focused through silica at the titanium–silica interface (see Fig. 1) into a Gaussian spot with the specific diameter of 70–270 µm measured at 1/e intensity level. The zone of the particle ejection was illuminated by a collimated beam of a CW laser (λ = 532 nm, P = 50 mW) parallel to the target surface. An objective formed a magnified image of the ejected material at the entrance of the intensifier of a gated CCD camera (Princeton Instruments Inc.). The picosecond laser was operating at 10 Hz. A laser pulse recorded by a photodiode (PD) was used to trigger a delay generator, which in turn controlled the camera gating and an electromechanical shutter. They were synchronized on the following pulse (that is to say, 100 ms later). The first pulse was blocked by the shutter, but the following pulse reached the target and initiated the material ejection. As the femtosecond laser was operating at 1 kHz, the single pulse was selected using an electrooptical device (Pockels cell). The delay generator triggered the Pockels cell and the ICCD camera with appropriate delays. Independently of the laser pulse width, the gate width in the ICCD camera was fixed at the level of 20 ns, while the delay of the camera gating after the laser pulse varied in the range of 0–20 µs. Each time-resolved photograph of the ejected material was obtained under irradiation of a fresh material area at the target. By varying the laser pulse energy, we referred the results of laser irradiation to the incident fluence calculated in the spot center. It is well known that nonlinear interaction of propagating intensive pulses with surrounding media can cause substantial modification of the initial beam profile [20]. According to [21], the critical power for self-focusing in fused silica reaches 4.3 MW for λ = 800 nm. This corresponds to the pulse energy of ∼2 µJ for the 450-fs pulses and 215 µJ for the 50-ps pulses. To reach the desirable maximum laser fluence (1–1.5 J/cm2 ), we varied the pulse energy up to 80 µJ for the 450-fs pulses and 1.2 mJ for the 50-ps pulses. The given maximum energies substantially exceed the estimated critical pulse energies for the self-focusing effect, especially in the case of 450-fs pulses (by a factor of 40).
Shadowgraphic imaging of laser transfer driven by metal film blistering
51
However, the resulting beam perturbation is limited in our case by the relatively small thickness (1.5 mm) of the silica support. To estimate the actual deviation of the energy distribution in the irradiated area from the expected Gaussian profile, we exposed the target with the thinnest metal layer to different pulse energies up to the mentioned maximum values. Afterwards, we measured the size of the metal-free round patterns occurring on the target surface. The laser fluence at the pattern boundary (Fp ) is given for a Gaussian beam by the simple formula 2
Q − Dp2 Fp = e 4w , πw 2
(1)
where Dp is the pattern diameter, w is Gaussian radius and Q is pulse energy. Assuming that Fp = const, the following relation between the experimentally measured values Dp and Q can be derived: Dp2 = 4w 2 ln Q − 4w 2 ln πw 2 Fp , (2) i.e.
Dp2 = A ln Q − B,
(3)
where A and B are constants. Indeed, we have found that expression (3) is valid for all examined irradiation conditions excluding only high-energy (Q > 60 µJ) 450-fs pulses. The revealed deviation from the relation (3) indicates deformation of the original Gaussian beam that makes impossible reliable calculation of the laser fluence in the spot center. Therefore, we have excluded the shadowgraphic data for the high-energy femtosecond pulses from the consideration below. The described verification procedure does not guarantee that the laser beam remains totally unperturbed below the mentioned energy level, but it gives a chance to avoid a substantial mistake in the fluence calculation.
3 Results The effect of laser irradiation with different pulse energies was preliminarily investigated for all available targets and pulse widths using optical microscopy. Observing the metallized side of the targets, we identified local ejection of the diamond layer and local removal of the titanium layer. As a result, we have determined two important parameters of the laser transfer process: (i) the minimum laser fluence required to eject diamond particles from the target surface, which has been defined as the transfer threshold (Ftr ), and (ii) the minimum laser fluence that allows local removal of the metal layer from the target, i.e. the metal removal threshold (Frem ). The obtained values are summarized in Fig. 2 as a function of the metal thickness (d). The metal removal threshold was found noticeably higher than the transfer threshold (approximately by a factor of two) for
Fig. 2 Threshold fluences for particle transfer and metal removal as functions of metal film thickness
all examined experimental conditions. Both thresholds are slightly lower for the femtosecond pulses. Within the laser fluence range limited by the two mentioned thresholds (i.e. Ftr < F < Frem ), only the diamond nanopowder is ejected from the target, while the underlying metal layer remains on it. The evident reason for the diamond ejection in this case is fast local displacement of the free surface of the metal film. Therefore, the given transfer mode was called a ‘blistering’ regime. Above the metal removal threshold, the diamond transfer results from the metal ejection. According to the modeling of the target heating discussed in more detail below, evaporation of the irradiated front surface of the metal film foregoes the observed metal removal. Under removal of thick metal layers, however, a substantial part of the ejected metal can remain in the solid state. Independently of the exact relation between different aggregative states in the removed metal, we define the given transfer mode (at F ≥ Frem ) as an ‘ablative’ regime. Observing with a microscope the irradiated target areas with the surviving metal film through the transparent support, we have revealed in some cases the appearance of sharply delineated round spots with uniformly increased optical reflectivity. This effect was interpreted as local exfoliation of the metal layer from the silica support. Indeed, reasoning from the refractive indices of titanium and silica (nTi = 1.86 + i · 2.56, nsi = 1.46 at λ = 540 nm), one can calculate that the Fresnel reflection of the silica/titanium boundary (R0 = 0.382) is less than the total reflection of two boundaries occurring under the metal exfoliation, namely silica/vacuum (R1 = 0.035) and vacuum/titanium (R2 = 0.495): R0 < R1 + R2 . The spot diameter increased with the incident fluence, but the reflectivity contrast remained unchanged. The gap between the metal film and the silica support became detectable only under cracking of the metal layer that often happened when the laser fluence approached the metal removal threshold.
52
T.V. Kononenko et al.
Fig. 3 Comparison of important thresholds for 50-ps pulses
The exfoliation phenomenon was observed only for the targets with thick titanium films (d ≥ 400 nm). The minimum values of the laser fluence that are necessary to cause the metal exfoliation in these targets by 50-ps pulses are depicted in Fig. 3 by crosses. The given exfoliation thresholds are compared in the figure with the transfer thresholds and the metal removal thresholds for the same pulse width. Besides, the minimum laser fluences able to initiate evaporation of the titanium surface contacting with silica are shown in Fig. 3 by a solid line. To obtain these data, we have simulated heating of the examined targets by 50-ps pulses by applying a numerical method described in detail elsewhere [8]. It is expected that the metal vaporization process begins after the local temperature reaches the boiling point and the material absorbs energy equal to the latent heat of vaporization. We have found that the calculated vaporization thresholds are very close to the experimental exfoliation thresholds when the last can be measured (i.e. for d ≥ 400 nm). Moreover, comparing the vaporization threshold with the metal removal threshold for the 50-nm-thick titanium film, one can conclude that the film is removed from the target immediately after beginning of the metal vaporization, so that observation of the exfoliation effect appears impossible. According to the simulation, the metal removal in this case is facilitated by through melting of the 50-nm-thick film. The thicker titanium films remain partially solid and can resist the vapor pressure produced at sufficiently high incident laser fluence. Thus, it seems established that the observed local exfoliation of the titanium film results from the metal vaporization and indicates occurrence of vapor-filled bubbles at the metal/silica interface. The area between two dashed lines drawn in Fig. 3 (i.e. Ftr (d) and Frem (d)) relates to those combinations of the metal thickness and the laser fluence which allow the laser transfer in the blistering regime. The vaporization threshold dependence (i.e. the solid line) divides this area into two regions. Below the solid line, the displacement of the
Fig. 4 Time-resolved photographs of ejected material after irradiation by 50-ps pulses: (a) diamond aggregates (delay 8 µs); (b) uniform plume (delay 300 ns)
free metal surface results from reversible expansion of the heated metal. The contribution of heated silica adjoining the metal is negligible. Based on the simulated temperature distribution in the target with a 50-nm-thick titanium film (F = 250 J/cm2 ) and using available data on thermal expansion of titanium [22] and silica [23], we have estimated that the free metal surface is shifted by ∼5 nm due to the titanium film expansion and only by ∼0.02 nm due to the silica expansion. For the thin-film target, the thermal expansion is the single mechanism responsible for the metal blistering and acceleration of the covering powder layer. Increase of the transfer threshold with the metal thickness means decreasing efficiency of the thermal expansion mechanism. As a result, for the titanium films thicker than ∼650 nm, the nanopowder ejection becomes impossible without formation of an expanding vapor-filled bubble at the metal/silica interface. In this case, local displacement of the whole metal film in the perpendicular direction is determined by the va-
Shadowgraphic imaging of laser transfer driven by metal film blistering
53
Fig. 5 Maximum propagation of ejected material as a function of time for a few laser fluences (τ = 50 ps, d = 400 nm)
por pressure dynamics. Existence of two competitive mechanisms of the material ejection must be taken into account in any theoretical consideration of the blistering regime. Experiments on shadowgraphic imaging of the ejected material were carried out in a wide range of laser fluences (from F = Ftr up to F Frem ) to study both possible transfer regimes. A few examples of the time-resolved photographs are shown in Fig. 4. The spatial resolution of the experimental setup was sufficient to distinguish only large (>5–10 µm) aggregates of the diamond nanoparticles (Fig. 4a). The time-resolved photographs revealed no qualitative changes in the ejection process under the transition from the blistering regime to the ablative regime. Rise of the applied laser fluence above the metal removal threshold facilitated increasing fragmentation of the ejected diamond layer and formation of a uniform plume (Fig. 4b). The propagation rate of the plume front in Fig. 4b reached 1.3 km/s that caused generation of a visible shock wave in the surrounding air. The maximum displacement of the ejected material from the target surface is depicted in Fig. 5 as a function of time delay after the laser pulse for a few applied laser fluences. The presented dependences are approximated by linear functions passing through the zero point. Hence, no deceleration of the ejected material takes place over all the controlled range of distances (0–700 µm). All calculated values of the ejection velocity are summarized in Fig. 6a and b separately for femtosecond and picosecond pulses. The ejection velocity demonstrates an approximately linear rise with the laser fluence, while the curve slope decreases for the targets with thicker metal films. The femtosecond pulses provide higher ejection velocities under the same conditions, but the difference is less than 50%. The maximum ejection velocity achieved in the experiments is 1.4 km/s (d = 50 nm, τ = 50 ps, F = 1.4 J/cm2 ) that is comparable with the highest reported data of other authors [5].
Fig. 6 Velocity of diamond aggregates ejected from different metal layers as a function of laser fluence: (a) for 450-fs pulses, (b) for 50-ps pulses (filled symbols indicate blistering regime). Insets show maximum ejection velocity in blistering regime as a function of fluence
Considering the transition from the blistering regime of the nanopowder ejection (filled symbols in Fig. 6) to the ablative regime (empty symbols), one can conclude that the boundary between them relates to an approximately constant level of the ejection velocity (∼100 m/s). As the space between the points in Fig. 6 is quite large, we have specified the maximum ejection velocities in the blistering regime via superposition of the metal removal thresholds presented in Fig. 2 on the linear trends shown by dashed lines in Fig. 6. The results are demonstrated in the insets of Fig. 6a and b: each point relates to a separate target and the metal thickness increases from left to right. For the femtosecond pulses, the maximum velocity in the blistering regime varies from 60 m/s to 180 m/s, but it remains practically constant (110– 120 m/s) for all targets in the case of the picosecond pulses. The last result looks quite surprising, as it means that the metal film is removed from any target immediately after the material ejection velocity reaches a certain value. Let us recall, however, that the conditions and mechanism of the metal removal change drastically under variation of the metal thickness. Indeed, removal of the thin titanium film (d = 50 nm) is the result of melt splashing, while the thick
54
film (d = 1700 nm) remains mainly solid and must be broken off over the perimeter of the irradiated area. The shadow photographs demonstrate highly directive propagation of the ejected material. The maximum divergence evaluated in our experiments was less than 5–10◦ for both the femtosecond and the picosecond pulses. This correlates well with the observations of other authors [14, 15] on the ablative transfer initiated by short laser pulses. However, our earlier experiments on micropatterning of diamond nanopowder in the blistering regime [7] have revealed strong transverse scattering of the transferred aggregates; the scattering angle increased with laser fluence up to 40◦ . The important distinction of the mentioned shadowgraphic experiments from the real laser patterning is the absence of a receiving substrate able to affect movement of the ejected material. It has been demonstrated by Sano et al. [24] that interaction of the receiving substrate with the laser-ablated metal plume propagating in air environment causes strong lateral expansion of the plume. The observed plume deformation results from resistance of the ambient gas compressed between the plume and the substrate. Noticeable transverse scattering of the nanopowder in our experiments [7] can be explained by similar reasons, namely by compression and turbulization of the ambient air resulting in lateral deviation of trajectories of the flying particles. Smaller transverse scattering can be achieved due to reduced ejection speed and the blistering regime is more suitable for such optimization than the ablative regime. Reduction of the ejection speed can be important also for survival of fragile biological materials (cells, bacteria, etc.) under laser transfer, as was noted in [5].
4 Conclusions Time-resolved shadowgraphic imaging was used to investigate local ejection of diamond nanopowder from a thin titanium film on a silica support under irradiation by 450-fs or 50-ps laser pulses. Ejection of the nanopowder in the blistering regime, i.e. without removal of the metal layer, was found possible for all examined targets in a limited fluence range. The ejection velocity measured in both the blistering and the ablative regimes has demonstrated a general trend of an approximately linear rise with laser fluence. The targets with thicker metal films demonstrated a smaller slope of the velocity dependence and a higher transfer threshold.
T.V. Kononenko et al.
The maximum ejection velocity that can be achieved in the blistering regime was evaluated as ∼100 m/s for all targets and pulse widths. Acknowledgements The authors would like to thank A.F. Popovich from General Physics Institute for participation in the target preparation. The work was supported partially by RFBR (grant no. 93106) and the Russian Ministry of Education and Science (grant nos. 02.740.11.0417 and P951).
References 1. J. Bohandy, B.F. Kim, F.J. Adrian, J. Appl. Phys. 60, 1538 (1986) 2. J.A. Barron, B.J. Spargo, B.R. Ringeisen, Appl. Phys. A 79, 1027 (2004) 3. P. Serra, J.M. Fernandez-Pradas, M. Colina, M. Duocastella, J. Domínguez, J.L. Morenza, J. Laser Micro/Nanoeng. 1, 236 (2006) 4. C. Boutopoulos, V. Tsouti, D. Goustouridis, S. Chatzandroulis, I. Zergioti, Appl. Phys. Lett. 93, 191109 (2008) 5. B. Hopp, T. Smausz, N. Barna, Cs. Vass, Zs. Antal, L. Kredics, D. Chrisey, J. Phys. D: Appl. Phys. 38, 833 (2005) 6. L. Rapp, C. Cibert, A.P. Alloncle, P. Delaporte, Appl. Surf. Sci. 255, 5439 (2009) 7. T.V. Kononenko, P. Alloncle, V.I. Konov, M. Sentis, Appl. Phys. A 94, 196 (2009) 8. T.V. Kononenko, I.A. Nagovitsyn, G.K. Chudinova, I.N. Mihailescu, Appl. Surf. Sci. 256, 2803 (2010) 9. V. Menezes, K. Takayama, T. Ohki, J. Gopalan, Appl. Phys. Lett. 87, 163504 (2005) 10. V. Menezes, K. Takayama, A. Gojani, S. Hosseini, Shock Waves 18, 393 (2008) 11. N.T. Kattamis, P.E. Purnick, R. Weiss, C.B. Arnold, Appl. Phys. Lett. 91, 171120 (2007) 12. A.B. Bullock, P.R. Bolton, J. Appl. Phys. 85, 460 (1999) 13. Y. Nakata, T. Okada, Appl. Phys. A 69, S275 (1999) 14. I. Zergioti, D.G. Papazoglou, A. Karaiskou, C. Fotakis, E. Gamaly, A. Rode, Appl. Surf. Sci. 208–209, 177 (2003) 15. D.G. Papazoglou, A. Karaiskou, I. Zergioti, C. Fotakis, Appl. Phys. Lett. 81, 1594 (2002) 16. D. Young, R.C.Y. Auyeung, A. Pique, D.B. Chrisey, D.D. Dlott, Appl. Phys. Lett. 78, 3169 (2001) 17. D. Young, R.C.Y. Auyeung, A. Pique, D.B. Chrisey, D.D. Dlott, Appl. Surf. Sci. 197–198, 181 (2002) 18. I. Zergioti, A. Karaiskou, D.G. Papazoglou, C. Fotakis, M. Kapsetaki, D. Kafetzopoulos, Appl. Surf. Sci. 247, 584 (2005) 19. P.S. DeCarli, J.C. Jamieson, Science 133, 1821 (1961) 20. A. Couairon, A. Mysyrowicz, Phys. Rep. 441(2–4), 47 (2007) 21. A. Brodeur, S.L. Chin, Phys. Rev. Lett. 80, 4406 (1998) 22. P.-F. Paradis, W.-K. Rhim, J. Chem. Thermodyn. 32, 123 (2000) 23. J. Ishi, T. Kimura, Metrologia 5, 50 (1969) 24. T. Sano, H. Yamada, T. Nakayama, I. Miyamoto, Appl. Surf. Sci. 186, 221 (2002)